0

Dynamic Systems and Control

2004;():1-6. doi:10.1115/IMECE2004-59173.

A robust control algorithm is developed for attitude control of an air spindle testbed based on variable structure control. The variable structure control approach is formulated for underactuated systems with actuated accelerations as control inputs. First order sliding surfaces are defined per actuated degrees of freedom as a linear combination of the tracking position and velocity errors of both actuated and unactuated coordinates. The controller law is determined based on Lyapunov theory and is shown to be asymptotically stable for the tiled air spindle axis where gravitational effects are included but only marginally stable for the vertical axis. Multiple step sliding surfaces are proposed to achieve asymptotic stability in the latter case. Simulations of the controller performance verify its effectiveness and accuracy.

Commentary by Dr. Valentin Fuster
2004;():7-14. doi:10.1115/IMECE2004-59327.

Hybrid electric vehicles provide promising alternatives to conventional engine-powered vehicles, offering improvements in fuel economy and emissions. Realization of these benefits depends, in part, upon proper control of the vehicle. This paper examines a variable structure control which switches between two operating points. The resulting sliding optimal control provides a better energy management strategy than that obtained conventionally from Pontryagin’s minimum principle. One of the operating points is zero engine power, thus the sliding optimal control is referred to as engine start-stop. Contrary to the general impression that the engine should stop at low speeds or during decelerations, new studies show that engine start-stop also improves fuel economy during highway cruising. The main contribution of this paper is to show that this “duty-cycle” operation mode is indeed optimal. The main tool used in the proof is Pontryagin’s minimum principle.

Commentary by Dr. Valentin Fuster
2004;():15-20. doi:10.1115/IMECE2004-59392.

We present here a method of deriving a reduced-order-model (ROM) from physical principles in the design of a mixed H∞ /LQG controller for general applications. The ROM has been derived using the Karhunen-Loeve decomposition and Galerkin’s procedure from the high-fidelity computational fluid dynamics (CFD) model. As an illustration, the mixed H∞ /LQG controller has been designed for an optical fiber draw process, where traditional designs relying on empirical lumped-parameter models for regulating the fiber diameter are less than optimal due to difficulties in making practical measurements in the furnace domain of the drawing process. Simulations show that the transient dynamics of the ROM closely matches those of the CFD model. The effects of the modeling errors on the robust control system have been studied numerically, which show that the robust controller effectively reduces the fiber-diameter fluctuations, and that the close-loop system is rather insensitive to different kinds of modeling errors.

Commentary by Dr. Valentin Fuster
2004;():21-28. doi:10.1115/IMECE2004-59470.

This paper treats end point regulation of multi-link light-weight flexible manipulators using the State Dependent Riccati Equation: SDRE method. It is well known that end point trajectory control using widely used feedback linearization techniques is not possible when the equilibrium state of the zero dynamics of the system is unstable or weakly stable. Furthermore, control saturation is a major problem in controlling nonlinear systems. In this paper, an optimal control problem is formulated for the derivation of control law with and without control constraints on the joint torques for a multi-link flexible manipulator and suboptimal control laws are designed using the SDRE method. For the purpose of control, pseudo joint angles and elastic modes of each link are regulated to their equilibrium values which correspond to the target end point. Weighting matrices in the quadratic performance index provide flexibility in shaping the pseudo angles and elastic modes trajectories. In the closed-loop system, the equilibrium state is asymptotically stable, and vibration is suppressed. Simulation results are presented for a two-link flexible manipulator, which show that in the closed-loop system, end point trajectory tracking is accomplished even with constraints on the control torque. Results also show that the transient characteristics of the pseudo angles and elastic modes can be easily shaped by the choice of the performance criterion.

Commentary by Dr. Valentin Fuster
2004;():29-36. doi:10.1115/IMECE2004-59532.

The problem addressed here is to determine controls for moving a load along specified trajectories which avoid obstacles. It is possible to use flat outputs to determine inputs when hoist motion is present. However, when hoist is locked, the system does not appear to be differentially flat, and hence the above approach could not be used. We propose an iterative algorithm for the problem of calculating trolley motions in this case. Results for load motions requiring (a) travel and traverse of the trolley and hoist, (b) travel and hoist, and (c) travel alone, are presented. We also use flat outputs to formulate the minimum time control problem as a nonlinear programming problem, with constraints arising from limits on trolley and hoist accelerations and velocities, and positive rope tension. Solutions obtained are also presented.

Commentary by Dr. Valentin Fuster
2004;():37-43. doi:10.1115/IMECE2004-59990.

Modern applications in robotics such as teleoperations and haptics require high performance force actuators. Pneumatic actuators have significant advantages over electrical motors in terms of force-to-mass ratio. However, position and force control of these actuators in applications that require high bandwidth is not trivial because of the compressibility of air and highly non-linear flow through pneumatic system components. In this paper, we develop a detailed model of a pneumatic actuator system comprised of a double acting cylinder and a proportional servo valve to be used in position, force or hybrid position and force control.

Commentary by Dr. Valentin Fuster
2004;():45-54. doi:10.1115/IMECE2004-60009.

In this paper a nonlinear model-based controller is designed to globally asymptotically stabilize a single-degree-of-freedom shape memory alloy (SMA) actuated manipulator. A three part model was constructed based on the dynamics/kinematics of the arm, the thermomechanical behavior of SMA’s, and an assumed heat transfer model consisting of electrical heating and natural convection. The backstepping control is used to calculate the applied voltage to the SMA wire. Initially, the SMA’s wire stress is assumed to be the control input of the system. The stress is then chosen to asymptotically stabilize the desired position. The applied voltage to the SMA wire is the actual control input. This voltage is calculated based on the desired stress and the SMA’s thermomechanical and heat transfer models. It is shown that the calculated voltage can globally asymptotically stabilize the system. Numerical simulations are performed to investigate stabilizing performance as well as other issues such as robustness. The results demonstrate that the backstepping controller designs is highly accurate in stabilization.

Topics: Manipulators
Commentary by Dr. Valentin Fuster
2004;():55-62. doi:10.1115/IMECE2004-60253.

This paper presents an optimal control design and experimental implementation for pointing and disturbance rejection in a laser steam steering system. The linear quadratic Gaussian (LQG) controller, which includes a stochastic disturbance model, as well as integral action, was designed and implemented to compensate for disturbances due to atmospheric turbulence in the optical path and mechanical vibration of the laser and optical components. The control design also considers the situation where the stochastic disturbances applied to the two beam axes are correlated and renders a multi-input-multi-output (2-by-2) output feedback controller. The experimental system consists of a two-axis tilt mirror driven by piezo-electric actuators for controlling the laser beam, a second actuated tilt mirror to generate disturbances, a position sensing device that senses the location of the beam on a target plane, and a real time computer for digital control. System identification is used to determine a state space model of the beam steering system for use in control system design. Experimental results are presented to demonstrate the effectiveness of the LQG optimal disturbance rejection for the prescribed stochastic disturbances.

Commentary by Dr. Valentin Fuster
2004;():63-69. doi:10.1115/IMECE2004-60255.

A fast converging Repetitive learning control scheme is proposed to track non circular piston profiles with sharp discontinuities. Modern piston profiles have sudden jumps and discontinuities which introduce large transient errors in tracking. This is handled by the learning control scheme built on top of the inner loop PD control. Simulations show that the learning scheme results in monotonic error convergence. The control law has an adjustable scalar which determines the convergence rate. The scheme involves filtering the control and tracking error values from the previous cycle to generate the learning control for the current cycle. The learning control algorithm generated a modified reference for the inner loop stabilizing feedback. We also show that Feedforward schemes based on stable plant inversion can be brought under this scheme by choosing appropriate learning filters. We also derive sufficiency conditions for stability of the proposed scheme.

Commentary by Dr. Valentin Fuster
2004;():71-80. doi:10.1115/IMECE2004-60256.

This paper presents the design and experimental implementation of an adaptive inverse control system for a two axis MEMS tilting mirror used for optical beam steering. The theoretical issues and practical design considerations involved in this task are discussed in detail. The first topic addressed is the system identification of input-output and state-space models of the MEMS mirror. Consistency among the following two system identification methods is verified: identification of a parameterized transfer function and identification of a state-space model by a subspace method. Next, a stabilizing feedback controller and an adaptive inverse control scheme based on an adaptive inverse QR recursive least-squares filter are developed. Finally, the experimental implementation of the control loops is described and the performance of the beam steering system is analyzed.

Commentary by Dr. Valentin Fuster
2004;():81-89. doi:10.1115/IMECE2004-61289.

This paper presents a control system design for a type of time-varying nonlinear system. The control system comprises neuro-fuzzy system identifier, Luenberger observer, backstepping controller and variable structure controller. We use adaptive neuro-fuzzy inference system to identify the plant in real time without the need of underlying mathematical model. However, some knowledge about the plant structure and upper bounds is required. With the use of observer, the control system can be designed from plant output and input alone while plant states are assumed unmeasurable. Controller is designed based on backstepping scheme and uncertainties from the plant identification and state estimation processes are handled by variable structure controller. Under some important assumptions, the control system is proved to be able to track a smooth desired trajectory with uniformly ultimately bounded tracking error. A simulation based on one-link flexible-joint robot manipulator is provided.

Commentary by Dr. Valentin Fuster
2004;():91-97. doi:10.1115/IMECE2004-61558.

This paper presents a novel linear time-varying (LTV) iterative learning control law that can provide additional performance while maintaining the robustness and convergence properties comparable to those obtained using traditional frequency domain design techniques. Design aspects of causal and non-causal linear time-invariant (LTI), along with the proposed LTV, ILC update laws are discussed and demonstrated using a simplified example. Asymptotic as well as monotonic convergence, robustness and performance characteristics of such systems are considered, and an equivalent condition to the frequency domain convergence condition is presented for the time-varying ILC. Lastly the ILC algorithm developed here is implemented on a Microscale Robotic Deposition system to provide experimental verification.

Commentary by Dr. Valentin Fuster
2004;():99-107. doi:10.1115/IMECE2004-61817.

Repetitive control has been used extensively for rejection of periodic disturbances, in systems that have to follow periodic trajectories. To date, most repetitive controllers have focused on rejection of additive periodic disturbances. This paper suggests the use of a repetitive control algorithm for rejection of periodic multiplicative disturbances. The first result is a simple design method of a new controller to reject the multiplicative disturbance effectively, provided that the period of the disturbance is known. This controller is based on the internal model principle and the design method consists of a simple norm condition. It is shown that this repetitive-type controller can reject the disturbance. The second result is an extension of the first one to the case that the period of the disturbance is unknown. A period estimator is added to the control system to identify the period of the multiplicative disturbance. The algorithm, consisting of an adaptive recursive least mean square method, is simple. It is shown that this adaptive controller can reject the disturbance with an uncertain period and guarantee the stability of the adaptive closed-loop system including the period estimator.

Commentary by Dr. Valentin Fuster
2004;():109-116. doi:10.1115/IMECE2004-61964.

Repetitive control system is a servo system that achieves zero steady-state tracking error for any periodic desired outputs and any periodic disturbance inputs with a known period. This paper deals with a repetitive control problem for the case where the plant has a norm-bounded uncertainty and the period of periodic disturbances is unknown but is within known lower and upper bounds. A new robust repetitive controller is proposed not only to guarantee the robust stability against the uncertainty but also to estimate the unknown period and to reject any periodic disturbances with the period. It is shown that the design problem is reduced to a simple and feasible l1 norm minimization problem. The controller can be obtained easily by solving a simple linear programming problem. Simulation results show the effectiveness of the proposed controller for rejection of periodic disturbances with uncertain periods.

Topics: Control systems
Commentary by Dr. Valentin Fuster
2004;():117-125. doi:10.1115/IMECE2004-59067.

This paper describes four conceptual designs of strong hybrid vehicle powertrains. These concepts enable conversion of conventional powertrains into strong hybrid powertrains with minimal tear-up to the existing architecture. These concepts are configured as follows: (1) incorporates an electric machine attached to the front axle of a conventional rear-wheel-drive vehicle; (2) a Flywheel-Alternator-Starter (FAS) system with a motor placed between the torque converter and the transmission; (3) same as previous one but where the torque converter is replaced by a starting clutch; and (4) a dual mode Electric Variable Transmission (EVT). These concepts provide extensive hybrid functionality such as, electric motor-only drive; launch assist, braking energy recovery and regeneration. Simulation results indicate that the proposed strong hybrid concepts have the potential to provide fuel economy gains of 19% to 26% over conventional powertrains.

Commentary by Dr. Valentin Fuster
2004;():127-136. doi:10.1115/IMECE2004-59079.

The paper deals with the simulation of the wall wetting dynamics in SI engines, making use of Recurrent Neural Networks (RNN). RNN are derived from the Multi Layer Feedforward Neural Networks, largely adopted for static mapping, by considering feedback connections between output and input layers. A Multi Input-Single Output structure has been adopted, assuming injected fuel, manifold pressure and engine speed as external input variables; the Air-Fuel Ratio at the exhaust gas oxygen sensor location has been considered as system output. The RNN has been trained (i.e. identified) and tested vs. a set of transient data measured on a commercial 4 cylinders SI engine at the test bench. The results show a good level of accuracy confirming the suitability of RNN for both HIL simulation or off-line identification of classical Mean Value Models with a drastic reduction of the calibration effort.

Commentary by Dr. Valentin Fuster
2004;():137-143. doi:10.1115/IMECE2004-59119.

The strut is one of the most important components in a vehicle suspension system. Since it is highly non-linear it is difficult to predict its performance characteristics using a physical mathematical model. However, neural networks have been successfully used as universal ‘black-box’ models in the identification and control of non-linear systems. This approach has been used to model a novel gas strut and the neural network was trained with experimental data obtained in the laboratory from simulated road profiles. The results obtained from the neural network demonstrated good agreement with the experimental results over a wide range of operation conditions. In contrast a linearised mathematical model using least square estimates of system parameters was shown to perform badly due to the highly non-linear nature of the system. A quarter car mathematical model was developed to predict strut behavior. It was shown that the two models produced different predictions of ride performance and it was argued that the neural network was preferable as it included the effects of non-linearities. Although the neural network model does not provide a good understanding of the physical behavior of the strut it is a useful tool for assessing vehicle ride and NVH performance due to its good computational efficiency and accuracy.

Commentary by Dr. Valentin Fuster
2004;():147-161. doi:10.1115/IMECE2004-59206.

This paper describes empirical investigations of the fluid field for a spool-type hydraulic valve with symmetrically distributed circular ports that is often found in an automotive VFS (Variable Force Solenoid) valve system. Through extensive data analysis, a general trend of fluid force and flow rate is derived as a function of pressure drop and valve opening. Aiming at further revealing the insights of the steady state spool valve fluid field, the equivalent jet angle and discharge coefficient are calculated from the measurements based on the lumped parameter models. New Non-Dimensional Artificial Neural Network (NDANN)-based hydraulic valve system models are also developed in this paper through the use of equivalent jet angle and discharge coefficient. By introducing the outputs of the new NDANN models into the lumped parameter model, fluid force and flow rate can be easily calculated. Therefore, the new approach calculates fluid force and flow rate as well as the intermediate variables (equivalent jet angle and discharge coefficient) with useful design implications. The network training and testing demonstrate that the NDANN fluid field estimators can accurately capture the relationship between the key geometry parameters and discharge coefficient/jet angle. The new approach also maintains the non-dimensional network configuration and possesses scalability with respect to the geometry parameters and key operating conditions. All these features make the new NDANN fluid field estimator a valuable tool for automotive hydraulic system design.

Commentary by Dr. Valentin Fuster
2004;():163-171. doi:10.1115/IMECE2004-59214.

Hybrid ground vehicles have motivated electric and steer-by-wire steering system technology due to restrictions on power source availability. Although these two steering systems are efficient, flexible, and environment friendly, the steer-by-wire system provides the opportunity for semi-autonomous and autonomous vehicle operation, as well as compliments a drive-by-wire architecture. For greater lateral vehicle performance, reduced maneuver transient time, and avoidance of undesirable vehicle motions through combined traction and steering control, a four wheel steering assembly with front and rear steering mechanism can uniformly control the wheels’ steering angle. In this paper, mathematical models will be developed for a front and rear rack and pinion steer-by-wire system. Accompanying linear and nonlinear controllers will be designed for operator commanded tracking by adjusting the three servo-motor assemblies. Representative numerical results are presented and discussed to support the evaluation of the four-wheel steering systems for sinusoidal and impulse-like steering maneuvers. The simulated vehicle four wheel steer-by-wire system results demonstrated better performance compared to the front steer-by-wire system.

Commentary by Dr. Valentin Fuster
2004;():173-180. doi:10.1115/IMECE2004-59592.

The Milliken Moment Method (MMM) is a way of efficiently organizing the results from a kinematically constrained vehicle test. Using the MMM and closed-loop dynamic simulation, the authors present a new method for evaluating vehicle handling, where the MMM assesses the overall performance capability of the vehicle, and the simulation determines how much of that performance is used for a closed-loop maneuver. By mapping the simulation data onto the yaw moment - lateral acceleration diagram from the MMM, this method allows a design engineer to explicitly quantify the tradeoff between the controllability and stability of a vehicle. Results from a design case study and comparison of the stability measure to phase plane analysis show that this method captures the tradeoff in design, and clearly represents the overall capability of the driver-vehicle system.

Commentary by Dr. Valentin Fuster
2004;():181-188. doi:10.1115/IMECE2004-59771.

In this paper, an adaptive model predictive control scheme is designed for speed control of heavy vehicles. The controller coordinates use of compression brakes and friction brakes on downhill slopes. Moreover the model predictive controller takes the actuator constraints into account. It is shown that accurate estimate of mass is necessary for safe and comfortable operation in closed-loop. Also knowledge of the road grade improves the results further by contributing in feedforward control. Therefore a recursive least square scheme with forgetting is used in parallel with the controller to update the estimates of mass and road grade. The mass and grade estimates converged to their actual values when rich excitations were provided. As a result the adaptation improved the closed-loop performance.

Commentary by Dr. Valentin Fuster
2004;():189-198. doi:10.1115/IMECE2004-59996.

Hybrid Electric Vehicles (HEVs) improvements in fuel economy and emissions strongly depend on the energy management strategy. Big obstacles to the control design are the model complexity and the necessity of “a priori” knowledge of torque and velocity profiles for optimal torque split. This paper presents and compares four different energy management approaches for the control of a parallel hybrid electric sport-utility-vehicle.

Commentary by Dr. Valentin Fuster
2004;():199-206. doi:10.1115/IMECE2004-60012.

Fault detection and isolation has become one of the most important aspects in vehicle control system design. In this paper, we present a technique for single sensor fault detection and isolation in automotive on-board applications. It combines model-based diagnostics and a qualitative modeling approach. The proposed method is appealing as it shifts the computational effort from on-line to off-line, making the algorithm suitable for low-cost real-time applications. The methodology can be cast in the framework of discrete-event fault diagnosis. A depth one transition relation algorithm for qualitative identification which guarantees completeness is developed and applied to a 3-degree-of-freedom (DOF) nonlinear vehicle model. The paper concludes with preliminary simulation results showing the effectiveness of the proposed scheme.

Topics: Sensors , Modeling , Vehicles
Commentary by Dr. Valentin Fuster
2004;():207-215. doi:10.1115/IMECE2004-60023.

The advent of integrated powertrain technologies offers great potential for improvements in efficiency and transient performance of powertrains. This work examines the integration of various powertrain subsystems onboard an earthmoving vehicle. The nature of the variable displacement pumping subsystem is similar in concept to other forms of continuously variable transmissions. Separate controllers are designed using gain scheduled H∞ synthesis to address the problems of efficiency and performance. Both controllers are combined through switching logic to enable explicit tradeoffs between the objectives and thus improve the powertrain productivity. The devised control strategies are implemented on a hardware-in-the-loop representation of an earthmoving vehicle powertrain. The baseline efficiency controller is further improved by a hybrid design that effectively deactivates part of the powertrain system when not in use.

Commentary by Dr. Valentin Fuster
2004;():217-224. doi:10.1115/IMECE2004-60029.

Maintaining proper membrane humidity is crucial to ensure optimal operation of a polymer electrolyte membrane fuel cell, which can be achieved by adding a humidifier to the fuel cell system. A membrane-type humidifier system using the fuel cell exhaust gas to humidify the dry inlet air is studied in this paper. We first develop a thermodynamic model, which captures the crucial dynamic variables of the humidifier, including pressure, flow, temperature and relative humidity of air flows. Steady state simulations are then conducted to optimize the humidifier design. Subsequently, dynamic simulations are performed to predict the behavior of the humidifier during typical transient operations for automotive applications. A simple proportional controller was designed to control the humidifier according to the fuel cell stack operating conditions.

Commentary by Dr. Valentin Fuster
2004;():225-232. doi:10.1115/IMECE2004-60376.

Estimation of the actual cylinder air charge and air-charge prediction several engine events into the future are needed to implement an air-fuel ratio feedforward controller. The estimator/predictor is usually based on the isothermal intake manifold model which assumes constant temperature of manifold air. However, fast thermocouple measurements have pointed to significant changes of manifold air temperature during typical tip-in/tip-out engine transients. In order to capture the temperature change effect, the manifold dynamics should be described by the so-called polytropic model. This paper presents an algebraic and simulation analyses of the influence of manifold modeling errors on the accuracy of air-charge prediction. The analysis is conducted for a typical predictor given in two variants depending on whether the throttle mass flow sensor or the manifold pressure sensor is used.

Commentary by Dr. Valentin Fuster
2004;():233-242. doi:10.1115/IMECE2004-60877.

Robotic vehicles are an attractive alternative to manned vehicles in hazardous or dangerous off road and urban environments. Present designs of robot vehicles employ wheels or tracks as the running gears and, in general, tracks provide superior mobility on rough or uneven terrain. This paper presents a multibody dynamics model of a tracked robotic vehicle for the purpose of predicting mobility in two different scenarios: 1) steep terrains, and 2) urban terrains in the form of staircases. In both scenarios we study the physical limitations on vehicle mobility imposed by key vehicle design variables and vehicle operating conditions. Example vehicle design variables include the location of the mass center, grouser penetration, and track/terrain friction. Example vehicle operating conditions include climbing under full versus partial track/terrain contact, and climbing on straight versus switch back courses.

Commentary by Dr. Valentin Fuster
2004;():243-248. doi:10.1115/IMECE2004-60942.

Variable valve timing (VVT) and variable compression ratio (VCR) are two technologies to obtain fuel economy benefit. On the other hand, there is a tradeoff among fuel economy, engine performance and emission levels. Advantages of two technologies vary a lot on different engine operating regions. Recently some experiments are conducted on a Port Fuel Injection (PFI) engine in a city drive cycle to investigate the fuel economy impact from VVT, VCR and the technology integration. The testing results show clearly that the synergy of two technologies has further improved the fuel economy, while suitable operating regions need to be determined where the maximal benefit can be achieved. A typical 1.8L four-cylinder gasoline engine is used for experiments using VVT and VCR technologies for fuel economy improvement. The objective is to create a synergy scheme for the optimal fuel economy performance. The supercharged testing engine with VVT and VCR can implement similar performance to that of a larger replacement engine. The fuel economy optimization problem is simply converted into searching for the lowest engine power output region with respect to the same fuel economy improvement level. These optimal points are useful to determine potential best fuel economy operating regions whether VVT and VCR should be implemented individually or combined together.

Commentary by Dr. Valentin Fuster
2004;():249-258. doi:10.1115/IMECE2004-61293.

The growing use of model-based engineering for powertrain design and validation imposes conflicting modeling requirements. Vehicle powertrain models must be accurate enough for design optimization, and fast enough for real-time verification on hardware-in-the-loop platforms. Because these two requirements conflict, one finds two families of powertrain models today: high-fidelity models and real-time models. Model surrogation can bridge the gap between these two families by extracting powertrain models accurate enough for optimization and fast enough for real time. This paper systematically derives surrogate torque converter, clutch friction, and clutch hydraulics models. Using these models, the simulation speeds for two automatic transmissions improve significantly without appreciable loss of accuracy.

Commentary by Dr. Valentin Fuster
2004;():259-266. doi:10.1115/IMECE2004-61436.

The SI engine load torque is key information for many engine and power train control systems. Since the torque is not measured in production vehicles, it needs to be estimated on-line. The paper presents design and analysis of second-order and third-order Luenberger load torque estimators. With the aim to reduce the estimator noise sensitivity without deteriorating its transient performance, an adaptive Kalman filter is proposed and compared with the Luenberger estimator. The adaptation mechanism is based on a load torque change detection algorithm. The estimators are examined by computer simulations and experiments.

Commentary by Dr. Valentin Fuster
2004;():267-274. doi:10.1115/IMECE2004-61638.

This paper presents an analysis and comparison of a vehicle with active front steering and rear-wheel steering. Based on linear analysis of base vehicle characteristics under varying speed and road surfaces, desirable vehicle response characteristics are presented and a set of performance matrices for active steering systems is formulated. Using pole-placement approach, controllability issues under active front wheel steering and rear- wheel steering controls are discussed. A frequency response optimization approach is then used to design the closed-loop controllers.

Commentary by Dr. Valentin Fuster
2004;():275-286. doi:10.1115/IMECE2004-61775.

System-level modeling and control strategy development for a hybrid fuel cell vehicle (HFCV) are presented in this paper. A reduced-order fuel cell model is created to accurately predict the fuel cell system efficiency while retaining dynamic effects of important reactant variables. The fuel cell system model is then integrated with a DC/DC converter, a Li-Ion battery, an electric drive and tire/vehicle dynamics to form a HFCV. The supervisory-level control problem of the HFCV is subsequently investigated. A stochastic dynamic programming (SDP) based approach is applied to obtain an optimal power management strategy. Simulations over different driving cycles showed that the SDP control strategy not only saved a significant amount of hydrogen but also produced smoother load for the fuel cell stack—both of which help the long term viability of the fuel cell technology for automotive applications.

Commentary by Dr. Valentin Fuster
2004;():287-296. doi:10.1115/IMECE2004-61851.

Six degrees of freedom nonlinear suspension roll model of a heavy truck axle unit is presented. The model includes the effects of nonlinear tire damping, road inclination angle, and variation of suspension angles. The roll model parametric analysis provides the influence of various geometric truck parameters that are important to the designer in stability viewpoint. Many of the important parameters such as nonlinear tire coefficients, road and suspension inclination angles, and some of their limiting values for stability are also discussed to facilitate the selection of proper actuators and sensors for automation. The dynamic stability of the axle unit is illustrated using parameters dependent phase portraits. These phase portraits provide qualitative information on truck’s directional stability.

Commentary by Dr. Valentin Fuster
2004;():297-306. doi:10.1115/IMECE2004-61955.

This paper presents a detailed analytical model development which can describe the dynamic behavior of the electromechanical brake-by-wire (BBW) system over the entire operating range. The complete model has 10 degree-of-freedom (DOF) and includes essential nonlinearities such as gear backlashes, Coulomb frictions, and disk gap clearance. Such a full model is reduced to 6 degree-of-freedom model in SIMULINK for simulation study of the effectiveness and the achievable performance of different hardware and controller designs, an invaluable tool in the early design stage of a product. Simulation results show that the model is able to reproduce various nonlinear characteristics including typical structural hysteresis as shown in real brake assembly. The linearized version of the full nonlinear model is then obtained for its modal properties to understand the modes that are critical to the low frequency dynamics of the overall system. The results of the modal analyses are subsequently utilized to obtain two simplified models, one for non-contact mode and the other for the contact mode of operation. The concepts of two simplified models well capture the dynamic characteristics of the system over the frequencies of interest and are being used in the controller (e.g., clamping force control) and estimator (e.g., gap clearance estimation) designs that are under investigation.

Commentary by Dr. Valentin Fuster
2004;():307-316. doi:10.1115/IMECE2004-61966.

Homogeneous charge compression ignition (HCCI) engines are a promising engine technology due to their low emissions and high efficiencies. Controlling the combustion timing is one of the significant challenges to practical HCCI engine implementations. In a spark-ignited engine, the combustion timing is controlled by the spark timing. In a Diesel engine, the timing of the direct fuel injection controls the combustion timing. HCCI engines lack such direct in-cylinder mechanisms. Many actuation methods for affecting the combustion timing have been proposed. These include intake air heating, variable valve timing, variable compression ratios, and exhaust throttling. On a multi-cylinder engine, the combustion timing may have to be adjusted on each cylinder independently. However, the cylinders are coupled through the intake and exhaust manifolds. For some of the proposed actuation methods, affecting the combustion timing on one cylinder influences the combustion timing of the other cylinders. In order to implement one of these actuation methods on a multi-cylinder engine, the engine controller must account for the cylinder-to-cylinder coupling effects. A multi-cylinder HCCI engine model for use in the control design process is presented. The model is comprehensive enough to capture the cylinder-to-cylinder coupling effects, yet simple enough for the rapid simulations required by the control design process. Although the model could be used for controller synthesis, the model is most useful as a starting point for generating a reduced-order model, or as a plant model for evaluating potential controllers. Specifically, the model includes the dynamics for affecting the combustion timing through exhaust throttling. The model is readily applicable to many of the other actuation methods, such as variable valve timing. Experimental results validating the model are also presented.

Commentary by Dr. Valentin Fuster
2004;():317-328. doi:10.1115/IMECE2004-61999.

A Continuously Variable Transmission (CVT) provides a continuum of gear ratios between desired limits. CVT is a promising automotive technology and a sundry of models has been researched to realize the potential benefits of a CVT. CVT, being a highly nonlinear system, has a definite operating regime where it is able to maximize the torque transmission. The numerical model presented in this paper is difficult to solve because of its sensitivity with respect to the initial operating conditions such as initial belt tension, axial forces, and driven preload. The present research focuses on using Genetic Algorithms (GA) to identify these operating conditions and to understand the various dynamic interactions in a metal pushing V-belt CVT. This paper uses continuous Coulomb friction approximation theory to model friction between the belt and the pulleys. The computational scheme, the mathematical models, and the results corresponding to different loading scenarios are discussed.

Commentary by Dr. Valentin Fuster
2004;():329-336. doi:10.1115/IMECE2004-62188.

Homogeneous charge compression ignition (HCCI) offers a promising way to improve efficiency and emissions. However, when HCCI is induced by reinducting exhaust gases, less power is produced. A possible solution is to couple HCCI with spark ignition (SI) operation at higher loads. This requires a way to smoothly switch between combustion modes. The authors present a multi-cycle, multi-mode combustion model to aid in understanding and controlling the mode transition. The model captures early ignition and low work after a switch from SI to HCCI. Furthermore, the model reveals a need to coordinate intake and exhaust valve timing to correct the HCCI phase and work. To demonstrate the model, the paper concludes with an example trajectory that maintains constant work and ignition phasing after a switch from SI to HCCI.

Commentary by Dr. Valentin Fuster
2004;():337-343. doi:10.1115/IMECE2004-59887.

Thermally induced bearing loads may cause serious problems for metal cutting spindles when used at high speeds. While proper spindle bearing setup can help minimize this problem, ultimately the plethora of factors that cause the loads to vary cannot be accounted for by the setup. Previous work has shown that bearing load control is possible, but that improved performance may be realized if the temperature environment of each bearing of a back-to-back pair is considered separately. The purpose of this paper is to provide the results of an experimental evaluation of a bearing load control strategy using two thermal actuators. A box spindle was modified and two electric heating tapes were placed around the front pair of back-to-back angular contact ball bearings, one around each bearing. Significant control over the bearing loads can be achieved for the conditions tested including some level of independent control. Unfortunately, there is cross-talk between the heat actuators and the bearing loads, and this interferes with the level of independent load control that can be achieved over the individual bearings. The significance of this issue is discussed and future research outlined. The authors conclude that it is possible to significantly reduce the transient load variation in both bearings with two rather than one actuator; additionally, the second control loop usually improves the steady-state behavior over that achievable with single heater placed over the rear bearing.

Commentary by Dr. Valentin Fuster
2004;():345-351. doi:10.1115/IMECE2004-61166.

Thermally induced bearing loads can potentially create serious problems for metal cutting spindles when used at high speeds. Proper spindle bearing set-up can minimize, but cannot eliminate this problem. Measuring the thermally induced load can alert the user to a potential problem or can be used to control the load directly. The purpose of this paper is to describe the design of a thermally induced bearing load sensor using strain gauges placed around the outer race. A box spindle with strain gauges on a pair of angular contact ball bearings located in the front of the spindle is used in this analysis. In order to calculate the thermally induced bearing load, the outputs of the strain gauges were recorded over a one second interval, sampled at 7500 Hz from each strain gauge and the root mean square of the deviation of this data from its mean is calculated. This value is a measure of the ball load. This calculated output is calibrated by a quadratic regression of these data to applied axial loads over a range of 0 to 2800 N. Values for each bearing were averaged to yield a front and rear value. The repeatability error for the front bearing sensor is 1.36%, and its accuracy is 98.9%. The repeatability error for the rear bearing sensor is 2.17% and its accuracy is 98.3%. These relatively low repeatability errors are attributable in part to filtering that does reduce the sensors’ bandwidth, but not significantly for measuring the relatively slowly changing thermally induced loads. Sensor design improvements and potential avenues of future research are discussed.

Commentary by Dr. Valentin Fuster
2004;():353-360. doi:10.1115/IMECE2004-60946.

Sandwiched deadbands can be seen in a wide variety of systems, such as electro-hydraulic systems controlled by closed-center valves. In such a system, the deadband is between the plant and actuator dynamics and therefore can not be compensated directly like an input deadband. Though this sandwiched deadband problem may be attenuated to certain degree through sophisticated advanced control techniques, the increased cost and the necessity of actuator state feedback prohibit their widespread application in the industry. An economical and popular method is to add an inverse deadband function in the controller to cancel or compensate the highly nonlinear behavior of the deadband. However, such a solution requires that the dynamics before the deadband (eg. the valve dynamics) is fast enough to be neglected — a requirement that can not be met in reality unless the closed loop bandwidth of the overall system is limited very low. To raise the achievable closed loop bandwidth for a much improved control performance, it is essential to be able to precisely characterize the effect of this sandwiched deadband on the stability and performance of the overall closed-loop system, which is the main focus of the paper. Specifically, a describing function based nonlinear analysis will be conducted to predict when the instability will occur and how the resulting limit cycle depends on the actuator dynamics and the targeted closed-loop bandwidth. Based on the analysis, the optimal closed-loop bandwidth can be determined to maximize the achievable overall system performance. The technique is applied to an electro-hydraulic system controlled by closed-center valves to optimize the controller design.

Topics: Valves
Commentary by Dr. Valentin Fuster
2004;():361-368. doi:10.1115/IMECE2004-61241.

In this paper, a nonlinear control scheme is developed for performing a cooperative task by hydraulic manipulators. The goal is to design a controller that allows two or more hydraulic robots to coordinately regulate an object’s position/orientation while maintaining desired internal forces on the object and sharing load. First the dynamic model of the whole system, including hydraulic functions, is derived. Then, a controller is designed, augmented by an on-line updating law to eliminate the steady-state error due to lack of knowledge about the weight of the object. Extended Lyapunov’s second method is used for stability analysis of the control system. The stability of the system is guaranteed by constructing a smooth Lyapunov function. Simulations are performed to substantiate the controller developed.

Commentary by Dr. Valentin Fuster
2004;():369-375. doi:10.1115/IMECE2004-61275.

The focus of this work is stabilization of hydraulic actuators during the transition from free motion to constraint motion and regulating the intermediate impacts that could drive the system unstable. In our past research, we introduced Lyapunov-based nonlinear control schemes capable of fulfilling the above goal by resting the implement on the surface of the environment before starting the sustained-contact motion. The hydraulic actuator’s stick-slip friction effect was, however, either not included in the analysis or not compensated by the control action. In this paper, the application of our previously introduced friction compensating position control scheme is extended to impact regulation of a hydraulic actuator. Theoretical solution and stability analyses as well as actual experiments prove that such control scheme is also effective for asymptotic impact control (with no position steady-state error) of hydraulic actuators in the presence of actuator’s dry friction.

Commentary by Dr. Valentin Fuster
2004;():377-384. doi:10.1115/IMECE2004-59442.

This paper describes the control of a direct-injection, liquid monoproplellant powered actuation system, which was developed for the purpose of providing mechanical power to autonomous human-scale robots. The actuation system utilizes the catalytic decomposition of a monopropllant as a hot gas generator for powering a pneumatic actuator. Pressurization of the respective sides of a pneumatic actuator. is provided via solenoid injection valves, which control the flow of the monopropellant through a catalyst pack into the respective sides of the cylinder. Depressurization is provided by a three-way proportional spool valve, which proportionally exhausts one of the two cylinder chambers. This paper describes a controller that coordinates the control of the two discrete injection valves, together with the control of the proportional exhaust valve, in order to provide actuator force tracking. The controller is implemented on a prototype system with a closed-loop servo controller. Experimental results indicate effective position and force tracking.

Commentary by Dr. Valentin Fuster
2004;():385-393. doi:10.1115/IMECE2004-60868.

This paper describes the design of and some preliminary control results for a hydraulically actuated human power amplifier. The system is in the form of an oar, with its reach and pitch degrees of freedom being hydraulically assisted. A robust PI force controller is proposed so that the hydraulic actuator force tracks a scaled copy of the force exerted by the human. Nonlinearities and uncertainties in the compression spring, as well as parametric uncertainties are taken into account. The passivity property of the closed loop system is also analyzed. The controller has been tested in simulations and experimentally. It is shown to be effective when pushing against an object, and in assisting in bearing static loads.

Topics: Design
Commentary by Dr. Valentin Fuster
2004;():395-404. doi:10.1115/IMECE2004-61244.

Human operated, hydraulic actuated machines are widely used in many high-power applications. Improving productivity, safety and task quality (eg. force feedback to the operator in a teleoperated scenario) has been the focus of past research. In addressing these issues, our research proposes and experimentally demonstrates a control technique that renders a hydraulic machine (teleoperated backhoe in this case) as a two-port, co-ordinated, passive machine. The passive teleoperated backhoe is driven by a human operator at one-port and interacts with the environment at the other. It guarantees interaction stability and safety with the human / work environment as the latter are usually passive. In previous work, a passive teleoperation algorithm was proposed for multi degree of freedom teleoperation of a hydraulic backhoe approximated by its kinematic behavior. The approximation led to severe performance deterioration under certain operating conditions. In this paper, a bondgraph based passive teleoperation architecture is proposed for the non-linear dynamic modeled backhoe. Passive control is designed in two stages. In the first stage, bondgraph based system inversion ideas are used to determine a coordination control law. In the second stage, a desired locked system (desired dynamics of the coordinated teleoperator) is defined and an appropriate control law is determined to ensure the passivity property of the locked teleoperator. The proposed passive control law is experimentally verified for its bilateral energy transfer ability and performance enhancements.

Commentary by Dr. Valentin Fuster
2004;():405-412. doi:10.1115/IMECE2004-61272.

This paper documents the development, theoretical analysis and experimental evaluation of a Lyapunov-based nonlinear control scheme for asymptotic force regulation of hydraulic actuators with friction. The complete discontinuous model of actuator friction, servo-valve dynamics, and nonlinear hydraulic functions are all included in the theoretical solution and stability analyses of the resulting nonsmooth system. The frictionless contact force is modeled as a linear stiffness. Filippov’s solution theory and the extension of LaSalle’s invariance principle to nonsmooth systems are employed to prove the asymptotic convergence of the system trajectories towards equilibria. Experiments complement the theoretical analysis in providing a solid foundation for implementation of the proposed control scheme for asymptotic force regulation of the hydraulic actuators despite friction effects.

Commentary by Dr. Valentin Fuster
2004;():413-420. doi:10.1115/IMECE2004-61626.

A method for synthesis of a robust adaptive scheme for a hydraulically driven manipulator, that takes full advantage of any known system dynamics to simplify the adaptive control problem for the unknown portion of the dynamics is presented. The control method is based on adaptive perturbation control. Using the Lyapunov approach, under slowly time-varying assumptions, it is shown that the tracking error and the parameter error remain bounded. This bound is a function of the ideal parameters and a bounded disturbance. The control algorithm decouples and linearizes the manipulator so that each joint behaves as an independent second-order system with fixed dynamics.

Commentary by Dr. Valentin Fuster
2004;():421-428. doi:10.1115/IMECE2004-59062.

Circulating fluidized beds (CFB) have been applied to a wide variety of chemical industry processes to reduce pollution and increase efficiency. Identifying the void fractions and the bed-height in the standpipe of the CFB is required for designing a controller to improve the overall system operation. An extended Kalman filter (EKF) algorithm has been applied in order to successfully estimate the states and the bed-height of the standpipe in the cold flow circulating fluidized bed (CFCFB) at the Department of Energy, National Energy Technology Laboratory. However, for some oscillating input cases, this method does not perform well. In addition, covariance matrices Q and R need to be assumed initially and depending upon initial conditions, for some cases, the EKF behaves unstably. In this research, a sliding mode estimator (SME) is applied in order to estimate the state, and the bed-height of the standpipe in the CFCFB. The sliding mode estimator requires the proper gain for tuning in order to have proper estimations. Test results show improvement in state estimation performance of SME over EKF.

Commentary by Dr. Valentin Fuster
2004;():429-433. doi:10.1115/IMECE2004-59352.

This paper describes a new control experiment developed for Mechanical Engineering undergraduate students. The experiment with a Shape Memory Alloy (SMA) actuated robotic arm is designed for the senior undergraduate laboratory (ME4006) in the Department of Mechanical Engineering at Virginia Tech. ME4006 is designed to provide the students with experience in experimental investigation of mechanical engineering systems. In designing this control experiment it was intended for the students to have a hands-on experiment with smart materials. Furthermore, students learn about control problems and limitations of theses materials along with sensing and actuation advantages of the SMAs. The experiment uses a problem solving approach; students are not given a procedure to follow for conducting the experiment. The problem is described in a memorandum to the students from a supervisor, who defines the purpose of the problem and defines the audience for the report.

Commentary by Dr. Valentin Fuster
2004;():435-442. doi:10.1115/IMECE2004-59372.

A small and fast reactor employing sodium-cooling system is expected and 4S (S uper S afe, S mall and S imple) Fast Reactor generating 50 or 10 Megawatts electric has been designed. This reactor pursues simple operation system, less maintenance, higher safety and improved economic features. The electric power is controlled by a neutron reflector, which is required to be elevated at a very low speed. In the present study, a reflector drive system by a magnetic impact type micro-stepping actuator has been proposed for the 4S reactor. The actuator is composed of electromagnets, air-core coils and internal inertia guided by linear slider. This internal inertia carries air-core coils, which are opposed to air-core coils arranged inside the actuator. The actuator adheres to a metal base wall and supports a payload by electromagnet adsorption force. The actuator moves by a magnetic repulsion force of internal inertia by adding momentary large electric current to opposed pair of air-core coils. A miniature model of “internal inertia” magnetic impact type actuator was made. Two different ways to support the internal inertia were proposed to restore the internal inertia to the original state. In this paper, experiment was performed to evaluate the basic characteristics of the two types of actuator. As a result, the actuator elevated about 4 micrometers per step in the case of the “electromagnet support type” and about 25 micrometers per step in the case of the “spring support type” with 10-kilogram payload. Also, it was found that the payload carries out minute quantity separation instantaneously with the actuator by the shock of a repulsion force. This separation decreases the magnetic force by which the payload pulls down the actuator and makes the actuator upward movement easy. Moreover, it became clear that the driving distance could be sorted by “effective magnetic impact impulse,” which can be calculated from current waveform. Therefore, driving distance can be controlled by condenser capacity and charged voltage that generates current waveform of the air-core coils. In light of these results, improvements will be made for design of an enlarged engineering model of the actuator. Experiments with larger payload are planned.

Topics: Actuators
Commentary by Dr. Valentin Fuster
2004;():443-449. doi:10.1115/IMECE2004-59472.

This paper describes a new robot capable of manipulating pelvic motion during human step training on a treadmill. The robot, PAM (Pelvic Assist Manipulator), uses two pneumatically actuated subsystems arranged in a tripod configuration to measure and control the pelvis of a person during body weight supported stepping on a treadmill. The device can be used in a back-drivable mode to record pelvic trajectories, either specified manually by a therapist or pre-recorded from unimpaired subjects, then replay these trajectories using a PD position feedback control law in the task space and a non-linear force control algorithm for each piston chamber. The control laws are presented, along with data that demonstrate the ability of the device to record and replay the pelvic motions that occur during normal walking.

Topics: Robots
Commentary by Dr. Valentin Fuster
2004;():451-456. doi:10.1115/IMECE2004-59568.

This paper presents a generic model for an integrated smart health monitoring system for infrastructures using multisensor fusion and condition assessment sheets. Though various techniques for health monitoring have been discussed extensively in the literature, little attention has been given to obtain high quality data from the measurement and sensing system by using an intrinsic knowledge base. The method proposed in this paper uses measurement data from different types of sensors with different resolutions and fuses it together based on the confidence in them derived from information not typically used in traditional data fusion methods. Examples of such information are operating temperature, frequency range, fatigue cycles, etc. These are fed as additional inputs to a fuzzy inference system (FIS) that has predefined membership functions for each of these variables. The outputs of the FIS are weights that are assigned to the different sensor measurement data that reflect the confidence in the sensor’s behavior and performance. A modular approach is adopted for the data fusion system. It allows adding or deleting a sensor, along with its fuzzy logic controller (FLC), anytime without affecting the entire data fusion system. The time history of problems and solutions taken to correct them are stored as a condition assessment sheet (CAS) that shows the health of each sensor and the entire measurement system at a glance. This work finds applications in the health management of civil infrastructures, power plants, airplanes and rocket/shuttle test facilities.

Commentary by Dr. Valentin Fuster
2004;():457-462. doi:10.1115/IMECE2004-59585.

The design of filters for the prediction of signals has been a widely studied field. For some of the applications, more accurate and further reaching algorithms are necessary. For example, in the prediction of irregular waves for wave energy converters, an accurate prediction of wave height and velocity are important in order to maximize the converter’s efficiency. This paper presents three prediction filters for such an application. The first algorithm is based on a simple autoregressive (AR) model and a standard least-squares estimation scheme. The second proposed filter is based on an autoregressive moving average (ARMA) model. The third filter is a fixed horizon predictive filter based on an AR structure, using a Genetic Algorithm to estimate its prediction parameters. All proposed algorithms are simulated using a Pierson-Moskowitz spectrum representing wind speeds of 30 knot.

Commentary by Dr. Valentin Fuster
2004;():463-471. doi:10.1115/IMECE2004-59607.

This paper shows the feasibility of using an intelligent systems approach to increase the accuracy of a 3D ultrasonic position estimation system for real-time image guided neurosurgery. Current image guided systems use camera based technology that is space-intensive, have an accuracy of about 1.0–2.0 mm, and are prone to occasional failures. The 3D system presented in this paper eliminates the space intensive camera, has an accuracy of around 1.0 mm in the operating range of about 200–400 mm, makes the system independent of line-of-sight occlusion problems, and is expected to pave the way for accurate fusion models of MRI and ultrasonography to account for brain shifts during surgery. Hence, the proposed system provides many more advantages over existing systems without compromising on the accuracy. This paper presents the system formulation, a neural network model that uses the raw signals, the electronic hardware for data acquisition and processing as well as simulation and actual results.

Commentary by Dr. Valentin Fuster
2004;():473-478. doi:10.1115/IMECE2004-59625.

This work presents the results of a mathematical modeling to study the dynamic behavior of a helical spring under a periodic excitation induced by a rotating cam. The spring is sleeved over a mandrel; thereby it is further subjected to a Coulomb damping force as it oscillates. Helical springs expand radially when they are compressed. The effect of this radial expansion is included in the mathematical model. Standard wave equation that includes variable Coulomb damping was used to examine the vibratory behavior of the spring. Numerical solution to the no-friction, constant-friction, and varying-friction forces were obtained from the wave equation, using Explicit Finite Difference method. Finite Element was used to model the radial expansion of the spring to determine the variations of the Coulomb friction force. The spring response to the prescribed cam excitation, under the variable Coulomb friction force, was found not to be significantly different from that of a previously assumed constant friction force, for the cases that were studied in this work. In case of postulating a variable damping force the residual vibrations of spring loops are slightly higher than of the constant damping force.

Commentary by Dr. Valentin Fuster
2004;():479-486. doi:10.1115/IMECE2004-60013.

The study intends to focus on the design of a thin-disc piezoceramic-driving ultrasonic actuator dedicated to discrete stepping motors. By the improved design and construction, the innovative ultrasonic actuator was developed and used as a stator. The electromechanical coupling characteristics based on the composite structure would produce the flexural wave on the stator, which consisted of a piezoceramic membrane bonded on a metal sheet. Due to the converse piezoelectric effect, the driving ability of the actuator came from the vibration of extension-shrinkage of a metal sheet corresponding to the frequency of a single-phase AC power. Under constraints at the specific geometry positions on the metal sheet, the varying behaviors of flexural waves were formed. The simple structure of an actuator demonstrated that the mechanical design of the actuator and the rotor could be separated, depending on what we need in pragmatic applications. And, its positioning accuracy could be reached through a closed loop servo control, i.e. Fuzzy Sliding Mode Control (FSMC). FSMC was used to automatically compensate nonlinearly mechanical behaviors such as dead-zone and hysteresis phenomena. Furthermore, FSMC scheme has successfully overcome the high frequency chattering phenomena with lower control effort while the motor is applied to position tracking, and also has been proven in the excellent robust ability for noise rejection.

Commentary by Dr. Valentin Fuster
2004;():487-493. doi:10.1115/IMECE2004-60025.

With the advent of Bluetooth and wireless 802.11 Ethernet protocols having transmission speeds up to 54 Mbps, wireless communication for closed-loop control is becoming more and more achievable. Some researchers have utilized Bluetooth networks for wireless control, resulting in successful stabilization of an unstable plant with a network controller. Previously, the authors of this paper developed a novel event-based control with time-based sensing and actuation communication method using 11 Mbps wireless Ethernet with the user datagram protocol (UDP). Near real-time control of an unstable Furuta pendulum with up to 250 Hz closed loop bandwidth was obtained using off the shelf hardware, Matlab, and Windows 2000 operating systems. The present work extends that communication scheme to two independent wireless loops that share a mutual goal, making additional communication between the two controllers advantageous. The communication framework for the coupled control in this ad hoc peer-to-peer network is presented along with some practical limitations. Data from a physical system implementing this framework demonstrates its effectiveness in application. The test plant couples a simple light tracking plant with a Furuta pendulum and a shared goal of maintaining line of sight (LOS) under normal conditions as well as reestablishing LOS in the case of lost contact due to sudden obstacles.

Topics: Networks
Commentary by Dr. Valentin Fuster
2004;():495-501. doi:10.1115/IMECE2004-60039.

Proposed in this paper is a new method to implement the high operating bandwidth sensor arrays. In certain control applications, it is necessary that a high bandwidth sensor be used to improve the efficiency of feedback. The design of a single sensor with the desired high bandwidth may not be easy and economically feasible. It is shown that the idea of sensor arrays can be utilized to obtain a cost effective and efficient solution to the problem posed. It is discussed that an effective data fusion scheme is necessary in order to implement the proposed sensor array that consists of low bandwidth pass-band sensors with possible overlapping operating regions. Moreover, we point out that obtaining accurate sensor models may not be always easy in practice and this may make the proposed sensor arrays inapplicable for certain applications. To address this issue, a new implementation scheme that utilizes feedback mechanisms to combine multi-sensor data is developed. The proposed framework is validated using simulation examples.

Commentary by Dr. Valentin Fuster
2004;():503-510. doi:10.1115/IMECE2004-60211.

The inputs to many ‘real’ mechanical systems are not readily measurable. For example, the input to the tire patch of the tires of automotive road vehicles is neither measurable nor easy to estimate. As conventional system identification procedures require input measurements or at least estimates of the inputs, a new approach for nonlinear system identification of mechanical systems, in the absence of an input measurement, is presented here. This approach uses a combination of time domain (Restoring Force) and frequency domain (Nonlinear Identification through Feedback of the Outputs (NIFO)) techniques. The time domain is used to characterize the nonlinearities in the system and the observed nonlinear characteristics are used in the frequency domain to build a model of the system. The method is applied to experimental tire-vehicle suspension system data.

Commentary by Dr. Valentin Fuster
2004;():511-519. doi:10.1115/IMECE2004-60424.

This paper describes the torque and magnetic flux analysis using an advanced dynamic dynamometer test bed for electromechanical motors. Test motor was tested under different levels of continuous loads and high bandwidth complex duty cycle loads in order to provide data to characterize the nonlinear properties of electric motors. The test bed to obtain torque saturation limit data is comprised of a servo motor which serves as a dynamic load emulator, a brake, a clutch, and full sensor array for comprehensive monitoring of test variables, (including magnetic flux density sensor). The magnetic flux test was conducted to discriminate magnetic saturation from torque saturation. Given the permanent magnet motor test system designed for a desired saturation limit, it is possible to predict saturation using a properly defined dynamic duty cycle norm as a function of the operating conditions and the actuator design parameters. In this paper, the temperature effects on the magnetic properties of the permanent magnet material were the principal objectives.

Topics: Torque , Engines
Commentary by Dr. Valentin Fuster
2004;():521-531. doi:10.1115/IMECE2004-60662.

In the recent decade, a great deal of research has been devoted to active control of the unsteady flow in a wide variety of components and/or subsystems of aircraft, automobile and marine vehicles and industrial fluid machinery, because small improvements in component and/or subsystem performance often lead to large payoffs. The term active flow control is used to describe the methods to actively manipulate flow fields with auxiliary power introduced to the flow. In this paper, a brief survey of the recent progress in active flow control research is made. The possibilities of further performance improvement using the theories and technologies of intelligent systems are discussed. Intelligent systems can be applied to improve sensing and actuating, increase model accuracy, and optimize the control schemes. The active flow control systems, on the other hand, may also challenge intelligent systems researchers and stimulate new development of intelligent tools.

Topics: Flow control
Commentary by Dr. Valentin Fuster
2004;():533-539. doi:10.1115/IMECE2004-60794.

Electro-active polymers like poly(vinylidene fluoride) — PVDF and their copolymers, have been emerging as a low cost substitutes for piezoelectric ceramics. Carbon nanotubes with their excellent physical and electronic properties can be used in these electro-active polymers to enhance their actuation and sensing properties. In the work presented here, actuators and sensors have been fabricated using piezoelectric polymers reinforced with carbon nanotubes. Single-walled and multi-walled nanotubes are used to reinforce the piezoelectric polymers. The response of these actuators and sensors are measured and the influence of the fabrication methods on the sensor performance is determined. The use of these materials as actuators and sensors to implement active vibration control is studied. These transducers also serve as additional damping materials added to the structure undergoing vibration due to their viscous damping properties. The change in properties of these piezoelectric polymers with different fabrication conditions and nanotube addition, though provokes doubts about standardization, invokes more research efforts on these new generation transducers.

Commentary by Dr. Valentin Fuster
2004;():541-546. doi:10.1115/IMECE2004-61478.

Turbine engines are frequently used in critical systems including the power plants and propulsion systems of aircrafts and ships. Frequent inspection and periodic maintenance have been necessary to ensure their proper functionality. Condition based maintenance of jet engines can significantly reduce their operational and maintenance costs, and, in the mean time, enhance safety and reliability. This study investigates the feasibility of establishing the utility of a dynamic network, i.e., projection network, to recognize hot air pass faults from measurements of e.g., fan speed, core speed, compressor inlet and exit temperatures and pressures, turbine exit temperatures, etc. Projection network is a nonlinear dynamic network architecture that provides stable oscillatory or non-oscillatory attractors. In contrast to the static mapping provided by e.g., neural networks and fuzzy systems, the projection network offers more functionality through its rich dynamics. When properly setup, its nonlinear dynamics can filter out noise from measurements, and classify/recognize complex patterns. This study established the utility of projection network for detection and diagnosis of several aircraft engine faults. This paper will also describe methods for both structure training and parameter tuning. Using these methods, projection networks were setup to recognize baseline, fan damage, high pressure turbine fault, and customer bleed valve fault.

Commentary by Dr. Valentin Fuster
2004;():547-553. doi:10.1115/IMECE2004-61733.

Investigation on using network for distributed systems is an important topic in the motion control industry. This paper presents solutions to time-delay and packet-loss problems encountered in distributed real-time operation of an open-loop unstable magnetic levitation (maglev) test bed via an Ethernet. A novel model predictive control strategy with optimal controller design is developed to overcome the adverse influences of time delays and packet losses. By using the prediction of system states and the event-driven and time-driven smart actuator simultaneously, the plant receives the current control signal in every sampling interval even at the presence of time delays and packet losses. Thus we can compensate the time-delay and packet-loss in a uniform way. The simulation and experimental results demonstrated the feasibility and effectiveness of this control algorithm for NCSs with long stochastic time delays and successive packet losses.

Commentary by Dr. Valentin Fuster
2004;():555-563. doi:10.1115/IMECE2004-61849.

In this paper, we develop a complete mathematical model of a shape memory alloy (SMA) wire actuated by electric power and a bias spring. The operation of the SMA actuator involves different physical phenomena, such as heat transfer, phase transformation with temperature hysteresis, stress–strain variations and electrical resistance variation accompanying the phase transformation. We model each of these phenomena in a modular fashion. A key feature of the proposed model is that one or more of its modules can be extended to fit other SMA applications. At the heart of the proposed model is a dynamic hysteresis model capable of representing minor hysteresis loops. We generate the temperature profile for the hysteresis model using lumped parameter analysis. We extend the variable sublayer model to represent actuator strain and electrical resistance. The dynamic properties of the hysteresis model are developed and are exploited in developing control system strategies. Control simulation case studies are presented.

Commentary by Dr. Valentin Fuster
2004;():565-574. doi:10.1115/IMECE2004-61961.

In this work the benefits of variable span morphing are considered, specifically with regards to the increased maneuverability of bank-to-turn cruise missiles. Along with an increase in range associated with variable span aircraft-in the case where both wings are equally extended-there is also the possibility of an additional advantage of higher control authority over both pitch and roll motions. In the case of roll control, one wing is extended while the other is contracted thus producing a moment about the longitudinal axis of the missile. Compared to conventional tail surface control, this moment can be substantial depending on the flight conditions. There is, however, some complexity involved in flight control of variable span aircraft; namely the shift of the missile’s center of mass and the dependence of the roll producing moment on the angle of attack. The following work attempts to address these complexities through nonlinear control. First, a full nonlinear model of the missile is presented that includes aerodynamic effects and changes in weight distribution. Nonlinear methods are then used to control the trajectory of the missile via the roll angle, angle of attack, and sideslip angle. The results show that, in comparison with conventional missile configurations, the addition of variable span morphing has the capability of increasing overall flight performance.

Topics: Missiles , Flight
Commentary by Dr. Valentin Fuster
2004;():575-580. doi:10.1115/IMECE2004-62006.

A new approach to the design and control of shape memory alloy (SMA) actuators, called Segmented Binary Control (SBC), is extended from previous work. The transient response of SBC is examined and is discovered to be inadequate in real time servo control because of significant latency times. A dramatic improvement is shown using a feedforward method in which a predetermined path is known and appropriate actions are calculated beforehand. In addition, this feedforward servo control of SMA is accomplished with only internal local feedback loops and no global feedback of position.

Commentary by Dr. Valentin Fuster
2004;():581-588. doi:10.1115/IMECE2004-62008.

The intensive use of simulation tools, such as simulink has made possible the development of embedded controllers using a plant model as a base. The performance in the simulation environment does not account however for the details of the hardware architecture. In this paper, a way to approximate some of the effects of the features of a particular microprocessor is derived for the Simulink environment. Worst-case execution time of the given control algorithm is performed as the first step. The computation error is also estimated for that particular algorithm. A block in the Simulink model could then express the computation latency due to hardware processing and the error due to computation accuracy. An automatic way of estimating these effects is presented here in order to achieve a higher integrability of software tools available for the designer. In particular, controller design involves deciding what hardware platform to choose for a particular application. Analysis and experimental results of a particular microprocessor are discussed.

Commentary by Dr. Valentin Fuster
2004;():589-597. doi:10.1115/IMECE2004-59066.

This paper presents a novel method for adaptive anti-swing fuzzy logic control for overhead cranes with hoisting. The control action is distributed between three fuzzy logic controllers (FLC’s): trolley controller, hoist controller, and anti-swing controller. A method for varying the ranges of the variables of the three controllers as a function of the crane’s parameters and/or motion variables is presented. Simulation examples show that the proposed controller can successfully drive overhead cranes under various operating conditions.

Commentary by Dr. Valentin Fuster
2004;():599-603. doi:10.1115/IMECE2004-59088.

Developing a control system for breathing devices to control the pressure mask is a task which involves the consideration of breath cycle fluctuations due to normal and/or abnormal conditions, variations in environmental conditions and the airflow time delays within the device. Using forward compensation an appropriate control scheme is developed to achieve this task. To investigate the dynamic response, the breathing device incorporating the control system is modeled using Simulink® in a Matlab® environment and some validation experiments are conducted. The proposed control scheme seems to meet the demand of the fluid flow time delay and compensate for system fluctuations and disturbances.

Topics: Pressure , Masks
Commentary by Dr. Valentin Fuster
2004;():605-611. doi:10.1115/IMECE2004-59334.

In this paper digital control solutions for speed control of direct current motor and level control of dual water tank system are described using conventional control algorithms such as (PID (Proportional + Integral + Derivative) and PI (Proportional + Integral)) as well as intelligent control algorithms based on fuzzy logic. Appropriate software tools are used to allow remote activation of the systems. Observation of the system behavior from remote terminals is made possible using novel tele-reality capability. This capability allows realistic CAD (Computer Aided Design) drawings that accurately represent the physical systems on the remote terminals to exhibit appropriate motion corresponding to the actual movement of the physical system in the University of Maryland Eastern Shore Mechatronics and Automation Laboratory (UMESMAL).

Commentary by Dr. Valentin Fuster
2004;():613-618. doi:10.1115/IMECE2004-59452.

Fin of a subsonic projectile produce maneuvering force and moment that control the rotation angle of a projectile fin during flight. The objective of this paper is to study the feasibility of using piezoelectric actuator and fuzzy logic control to create a smart fin. The fin is rotated by a beam-based piezoelectric actuator, which has an end fixed to the rotation axle of the fine while the other end is pinned at the tip of the fin. A model of the dynamics of the system is obtained using the finite element approach. A fuzzy logic controller for the fin is designed. The membership functions of the fuzzy variables for this controller are determined using a hybrid genetic algorithm. Simulation results show that the proposed controller worked satisfactorily.

Commentary by Dr. Valentin Fuster
2004;():619-624. doi:10.1115/IMECE2004-59549.

Foot Ulcer in diabetic patients is a serious medical problem. A major contributor for the development of diabetic foot ulcers is a high, localized plantar foot pressure. It is believed that in diabetes the nerves in the extreme parts of the human body are damaged and cause deregulated blood flow, which may cause an insufficient blood supply. This can lead to a loss of feeling, change in shape of the feet, necrosis and ulcerations, and ultimately to partial or total amputation of the body part. The loss of feeling in the feet results in a loss of feedback to control the foot pressure distribution. It is proposed that high foot pressure concentration can be avoided by using an active, intelligent shoe insert, which is based on the mechanics of smart materials. This paper investigates the controls schemes necessary to accomplish an external foot pressure distribution scheme for preventing ulcerations or the progression of existing ulcers. A simple mathematical model of the shoe insert is developed. Foot pressure distributions for healthy subjects are used as a basis to control elevated foot pressures by changing the shape of the shoe insert. The optimal shape of the shoe insert with regard to the existing pressure distribution is computed. The optimal shape is implemented using different control schemes. The performance and the efficiency of the proposed control schemes are compared and analyzed. The main advantage of the proposed active shoe insert is its capability to sense the pressure peaks, change the pressure distribution, and provide stimuli for increased blood flow in the diabetic feet. [1,2,3]

Commentary by Dr. Valentin Fuster
2004;():625-634. doi:10.1115/IMECE2004-59589.

Design of an efficient fuzzy logic controller involves the optimization of parameters of fuzzy sets and proper choice of rule base. There are several techniques reported in recent literature that use neural network architecture and genetic algorithms to learn and optimize a fuzzy logic controller. This paper presents methodologies to learn and optimize fuzzy logic controller parameters that use learning capabilities of neural network. Concepts of model predictive control (MPC) have been used to obtain optimal signal to train the neural network via backpropagation. The strategies developed have been applied to control an inverted pendulum and results have been compared for two different fuzzy logic controllers developed with the help of neural networks. The first neural network emulates a PD controller, while the second controller is developed based on MPC. The proposed approach can be applied to learn fuzzy logic controller parameter online via the use of dynamic backpropagation. The results show that the Neuro-Fuzzy approaches were able to learn rule base and identify membership function parameters accurately.

Commentary by Dr. Valentin Fuster
2004;():635-644. doi:10.1115/IMECE2004-59627.

In this paper, two approaches for obstacle detection and position estimation are presented. One is an algorithm based on M out of N detection logic and the other is an algorithm based on the probabilistic Interacting Multiple Model (IMM) method. The M out of N threshold-based algorithm declares that there is an obstacle present if it gets M validated measurements out of N consecutive measurements. IMM algorithm runs two different models in parallel, each based on a different hypothesis. One model assumes that there is an obstacle present while the other model assumes that there is no obstacle present in the sensor field of view. The performances of the two algorithms are compared based on their false alarm rate and detection speed. At first, Monte Carlo simulations are performed using only the false measurements to determine the thresholds for each method that generate a similar number of false detections. Using these thresholds, the detection speed of each method is compared and it is shown that the IMM-based algorithm is superior to the M out of N logic-based algorithm.

Topics: Algorithms
Commentary by Dr. Valentin Fuster
2004;():645-652. doi:10.1115/IMECE2004-59742.

A game theoretic based scheme is considered in this study for multidisciplinary design optimization under uncertain conditions. The methodology developed is illustrated by considering the example of an internal combustion (IC) engine. Various game protocols are used to model the optimization process and the results obtained are compared with each other. A genetic algorithm (GA) is used as an optimization and constraining tool. Convergence, constraint handling and processing time are considered to evaluate the efficacy of the methodology developed.

Commentary by Dr. Valentin Fuster
2004;():653-661. doi:10.1115/IMECE2004-59844.

Using neural networks, this paper proposes a new model-following adaptive control design technique for nonlinear systems. The nonlinear system for which the method is applicable is assumed to be of known order. Furthermore, it is assumed that using a nominal model an appropriate nominal controller has been designed for the system. However, it is well-known that because of unmodeled dynamics and/or parameter uncertainties, a nominal controller seldom works the way it is intended to; and sometimes it even leads to instability. Hence there is a need to modify this nominal controller online, in a stable manner, to suppress these unwanted behaviors. An online control adaptation procedure proposed in this paper to achieve this objective. The control design is carried out in two steps: (i) synthesis of a set of neural networks which collectively capture the algebraic function that arises either because of the unmodeled dynamics or uncertainties in parameters and (ii) computation of a controller that drives the state of the actual plant to that of a desired nominal model. The neural network weight update rule is derived using Lyapunov theory, which guarantees both stability of the error dynamics as well as boundedness of the weights of the neural networks. Unlike existing methods, a distinct characteristic of the adaptation procedure presented in this paper is that it is independent of the technique used to design the nominal controller; and hence can be used in conjunction with any known control design technique. Moreover, this technique is applicable to non-square and non-affine systems as well. Numerical results for a fairly-challenging problem are presented in this paper, which demonstrate these features and clearly bring out the potential of the proposed approach.

Topics: Design
Commentary by Dr. Valentin Fuster
2004;():663-670. doi:10.1115/IMECE2004-60071.

This paper investigates the architectural effect on adaptive neuro-fuzzy inference system (ANFIS) for machine condition assessment. The study was motivated by ANFIS’s limitation in adapting its architecture to map the modeled input output relationship. Based on the grid input space partition method, two elements in defining an ANFIS architecture were studied: the type of the membership function (MF) and the MF number assigned to ANFIS inputs. A new modeling accuracy index was introduced to address the limitation of the traditional root mean square error (RMSE) in describing the effect of the MF type. The analysis showed that wide core membership functions enabled a smaller RMSE than narrow core membership functions for machine defect severity classification. It is further shown that selecting appropriate MF number is critical to ensuring accuracy of ANFIS, considering the overfitting problem. These results were experimentally investigated on a bearing test bed, where defect severity classification and dynamic load estimation were evaluated. The experiments agreed well with the theoretical analysis.

Topics: Machinery
Commentary by Dr. Valentin Fuster
2004;():671-677. doi:10.1115/IMECE2004-60113.

Over the past few years, much research has been performed on understanding the dynamics of an ultra-large, flexible toroidal satellite component subject to an internal pressure. However, the harsh environment of space is no place for inflated, membrane-like materials for fear of micro meteorite bombardment and subsequent puncture. Addressing this issue directly, United Applied Technologies (Huntsville, AL) has developed a novel, thin film casting approach to create a self-rigidizing torus. Once inflated, the torus structure is able to support its own shape, thus eliminating the need for any internal pressure. The self-rigidizing torus is extremely flexible, much more so than its pressurized predecessors. Such compliancy makes modal testing extremely difficult. However, through careful application of traditional modal testing techniques (shaker and accelerometer testing), the damped natural frequencies and mode shapes of the self-rigidizing torus can be discerned in the frequencies and mode shapes of the self-rigidizing torus can be discerned in the frequency bandwidth of interest, 1–12 Hz.

Topics: Satellites
Commentary by Dr. Valentin Fuster
2004;():679-685. doi:10.1115/IMECE2004-60204.

A novel method, independent component analysis (ICA), is introduced to gas metal arc welding (GMAW) process monitoring. ICA was applied to arc signals, i.e. welding current and arc voltage, to remove the correlation between them and extract an independent component (IC). Two series of robotic GMAW experiments were carried out to study the feasibility of ICA for online monitoring. It was found that IC put up an abnormity when there was a step disturbance in the welding process. Experimental results showed that the IC could be used as a state variable representing the process variation. By applying statistical process control (SPC) for the obtained IC, a burning-through defect was isolated from the normal operation. The comparison between ICA and PCA was also made for the processes, which led an interesting result and was in need for further study.

Commentary by Dr. Valentin Fuster
2004;():687-696. doi:10.1115/IMECE2004-60238.

Neural network is a powerful tool that can be utilized for structural damage detection and health monitoring. Since damage usually varies/reduces stiffness, frequency response variation can be used as indicator for damage occurrence. A well designed neural network can correlate frequency response variation to damage localization/severity without resorting to detailed structural modeling. While various neural network based approaches have been developed, their effectiveness, efficiency, and robustness oftentimes rely on the selection of several important parameters in the network construction. One of the key performance metrics for a neural network is the learning rate. Although the dynamic steepest descent algorithm (DSD) and fuzzy steepest descent algorithm (FSD) have shown promising possibility of improving the learning convergence speed significantly without increasing the computational effort, its performance still depends on the selection of control parameters and control strategy. In this paper, a tunable steepest descent algorithm (TSD) improving the performance of the dynamic steepest descent algorithm is proposed. A numerical benchmark example shows that the proposed algorithm significantly improves the convergence rates of the backpropagation algorithm. A structural health monitoring system incorporated with the neural network trained by the adaptive learning algorithm is developed for detecting the impact damage.

Commentary by Dr. Valentin Fuster
2004;():697-705. doi:10.1115/IMECE2004-60502.

This paper deals with the design and optimization of a Fuzzy Logic Controller that is used in the obstacle avoidance and path tracking problems of mobile robot navigation. The Fuzzy Logic controller is tuned using reinforcement learning controlled Genetic Algorithm. The operator probabilities of the Genetic Algorithm are adapted using reinforcement learning technique. The reinforcement learning algorithm used in this paper is Q-learning, a recently developed reinforcement learning algorithm. The performance of the Fuzzy-Logic Controller tuned with reinforcement controlled Genetic Algorithm is then compared with the one tuned with uncontrolled Genetic Algorithm. The theory is applied to a two-wheeled mobile robot’s path tracking problem. It is shown that the performance of the Fuzzy-Logic controller tuned by Genetic Algorithm controlled via reinforcement learning is better than the performance of the Fuzzy-Logic controller tuned via uncontrolled Genetic Algorithm.

Commentary by Dr. Valentin Fuster
2004;():707-713. doi:10.1115/IMECE2004-60800.

This paper describes the development of the utility of a dynamic neural network known as projection network for pattern classification. It first gives the derivation of the projection network, and then describes the network architecture and analyzes properties such as equilibrium points and their stability condition. The procedures for utilizing the projection network for pattern classification problem are established and the benefits are discussed. The proposed classification system is then tested with well-known benchmark data sets, namely the Fisher’s iris data, the heart disease data and the credit screening data and the results are compared to other classifiers including Neural Network Rule Base (NNRB), Genetic Algorithm Rule Base (GARB), Rough Set, and C4.5 decision tree.

Topics: Networks
Commentary by Dr. Valentin Fuster
2004;():717-726. doi:10.1115/IMECE2004-60870.

A coal preparation plant typically has multiple cleaning circuits based on size of coal particles. The traditional way of optimizing the plant output and meeting the product constraints such as ash, sulfur and moisture content is to equalize the average product quality from each circuit. The present study includes multiple incremental product quality approach to optimize the clean coal recovery while satisfying the product constraints. The plant output was optimized at the given constraints of 7.5% ash and 1.3% sulfur. It was observed that utilizing incremental product quality process gives 2.13% higher yield which can generate additional revenue of $4,260,000 per annum than that obtained by using the equal average product quality approach in this particular case. This paper introduces a novel approach for optimizing plant output using Genetic Algorithms (GA) while satisfying the multiple quality constraints. The same plant product constraints were used for GA based analysis. The results showed that using GA as an optimization process gives 2.23% higher yield that will result in additional revenue generation of $4,460,000 per annum than average product quality approach. The GA serves as an alternative process to optimize the coal processing plant yield with multiple quality constraints.

Commentary by Dr. Valentin Fuster
2004;():727-732. doi:10.1115/IMECE2004-60964.

Miniaturized, distributed, networked sensors — called motes — promise to be smaller, less expensive and more versatile than other sensing alternatives. While these motes may have less individual reliability, high accuracy for the overall system is still desirable. Sensor validation and fusion algorithms provide a mechanism to extract pertinent information from massively sensed data and identify incipient sensor failures. Fuzzy approaches have proven to be effective and robust in challenging sensor validation and fusion applications. The algorithm developed in this paper — called mote-FVF (fuzzy validation and fusion) — uses a fuzzy approach to define the correlation among sensor readings, assign a confidence value to each of them, and perform a fused weighted average. A sensor network implementing mote-FVF for monitoring the illuminance in a dimmable fluorescent lighting environment empirically demonstrates the timely response of the algorithm to sudden changes in normal operating conditions while correctly isolating faulty sensor readings.

Commentary by Dr. Valentin Fuster
2004;():733-742. doi:10.1115/IMECE2004-61317.

This paper presents a dynamic reconfigurable control strategy based on the Direct Machining And Control (DMAC) research at Brigham Young University. We propose a reconfigurable framework which will allow a DMAC compliant machine to be controlled by a variety of applications and control laws. This Reconfigurable Mechanism for Application Control (RMAC) paradigm uses a hierarchical architecture to configure a mechanism into a device driver for direct control by an application like CAD/CAM. The paradigm is one of a mechanism device driver assigned to each mechanism class or model, and uses only the master model to control the mechanism. The traditional M&G code language is no longer necessary since motion entities (NURBS, lines, arcs, etc) are passed directly to the mechanism. The design strategy of using dynamic-link libraries (DLL) to form a mechanism device driver permits a mechanism to assume different operating configurations, depending on the number of axes and machine resolution. For example, the machine can perform as a material removal machine in one instant, and then, by loading a new device driver, act as a Coordinate Measuring Machine (CMM). This strategy is possible because DMAC is a software and networked-based control architecture. Both the CAD/CAM planning software and the real-time control software reside on the same PC. The CAM process plan can thus directly control the machine without need for process plan decomposition into the forms supported by the native controller. The architectural framework is explained in detail and the methodology for control software reconfiguration into a device driver is presented. For demonstration purposes three device drivers will be implemented on one machine to demonstrate feasibility and usefulness.

Commentary by Dr. Valentin Fuster
2004;():743-748. doi:10.1115/IMECE2004-61924.

In this paper, we present a case study of several examples involving optimal positive real control design for positive real plants. Specifically, we apply genetic algorithms to obtain low order optimal control designs with and without the strict positive real constraint on the compensator. This study shows that genetic search strategy is an effective approach for designing low order optimal controllers with complex constraints such as strict positive realness of the compensator. Furthermore, the examples studied also shed some new light on optimal positive real control problems.

Commentary by Dr. Valentin Fuster
2004;():749-757. doi:10.1115/IMECE2004-59212.

In this paper we discuss an end-point detection (EPD) method for the dielectric linear chemical-mechanical planarization (CMP) processes. The proposed EPD algorithms utilize the interferometry optical signals to determine the film post-thicknesses. A set of collected broadband spectral signals is formed as an spectral image. An image-matching technique is then used to match the pre-processed signal image to the reference image template obtained at the target film thickness. Several matching criteria are discussed and compared. We find that the image correlation coefficient is a good indicator to determine the process end-point. We also consider the impact of the material removal rate variations on the interferometry spectral signals. An analytical calculation is carried out to find an extraction and compression searching range of the spectral image to compensate for the removal rate uncertainties in real processes. The correctness and effectiveness of the proposed algorithms have been demonstrated through applications to an inter-metal dielectric (IMD) device CMP process. Compared with other optical EPD methods, the proposed image-matching method are robust to the CMP process variations.

Commentary by Dr. Valentin Fuster
2004;():759-767. doi:10.1115/IMECE2004-59546.

Process Control is one of the key methods to improve manufacturing quality. This research proposes a neural network based run-to-run process control scheme that is adaptive to the time-varying environment. Two multilayer feedforward neural networks are implemented to conduct the process control and system identification duties. The controller neural network equips the control system with more capability in handling complicate nonlinear processes. With the system information provided by this neural network, batch polishing time (T) an additional control variable, can be implemented along with the commonly used down force (p) and relative speed between the plashing pad and the plashed wafer (v). Computer simulations and experiments on copper chemical mechanical polishing processes illustrate that in drafting suppression and environmental changing adaptation that the proposed neural network based run-to-run controller (NNRTRC) performs better than the double exponentially weighted moving average (d-EWMA) approach. It is also suggested that the proposed approach can be further implemented as both an end-point detector and a pad-conditioning sensor.

Commentary by Dr. Valentin Fuster
2004;():769-783. doi:10.1115/IMECE2004-60266.

This paper presents simulation and implementation of a sliding mode controller on a pneumatic servo system. Typical controllers for pneumatic systems are fixed gain linear controllers. This study incorporates the non-linear characteristics of the pneumatic actuator into the controller design. A detailed mathematical model of a single ended pneumatic cylinder driven by a servo valve is developed. Effects of non-linear air flow through the valve, compressibility in the cylinder chambers and the residual volume of the connecting pipes are carefully considered. Both viscous and Coulomb frictions within the cylinder have been taken into account. A position tracking sliding mode control strategy is developed and implemented. A series of experiments and simulations are performed to show that the control performance is satisfactory in both tracking and regulation.

Commentary by Dr. Valentin Fuster
2004;():785-793. doi:10.1115/IMECE2004-59224.

In this study, the steady-state vibration response of a gearbox with gear tooth faults is investigated. Based on the analytical expression of the position-dependent mesh stiffness of the gear with perfect gear teeth derived with the potential energy method and the characteristics of involute gear teeth, expressions of the mesh stiffness of a gear with tooth faults such as tooth chip, tooth crack, and tooth breakage are derived. Using a coupled lateral and torsional vibration model of a one-stage spur gear pair, we have numerically solved a set of nonlinear equations and obtained typical vibration response diagrams of the gear pair with perfect gears and gears with tooth faults. This study reveals the relationship between the waveforms of the vibration and the types of local faults of the gear. These results are useful for identification of vibration signatures when there are these types of tooth faults.

Commentary by Dr. Valentin Fuster
2004;():795-802. doi:10.1115/IMECE2004-59450.

Track tension is closely related to the maneuverability of tracked vehicles and the durability of their tracks and suspension systems. The tension needs to be maintained at an optimum level throughout the maneuver in order to minimize the excessive load on the tracks and to prevent track peel-off from the sprocket. In this paper, a track tension control system is developed for tracked vehicles which are subject to various maneuvering tasks. It consists of track tension estimator, track tension controller and hydraulic unit. The tension around the idler and sprocket is estimated in real-time, respectively. Using the estimated track tension and considering the highly nonlinear vehicle characteristics, a fuzzy logic controller is designed in order to control the track tension in the vehicles. The performance of the proposed tension control system is verified through experimental field tests.

Commentary by Dr. Valentin Fuster
2004;():803-811. doi:10.1115/IMECE2004-59516.

To cope with the growing demands for simulation models of ever increasing complex industrial systems, the research community effort has been mainly focused in creating different software tools which simplify the modeling process. This work describes how the Object-Oriented Modeling language PML (Physical Modeling Language) automates the modeling process by using physical knowledge in order to set the mathematical model of the system. PML introduces new modular structures to represent the physical knowledge required to model a system, making a clear separation between the physical behaviour representation (declarative knowledge) and the computational aspects of model simulation (procedural knowledge).

Topics: Modeling
Commentary by Dr. Valentin Fuster
2004;():813-820. doi:10.1115/IMECE2004-59591.

This paper uses an air conditioning system to illustrate the benefits of iteratively combining first principles and system identification techniques to develop control-oriented models of complex systems. A transcritical vapor compression system is initially modeled with first principles and then verified with experimental data. Both SISO and MIMO system identification techniques are then used to construct locally linear models. Motivated by the ability to capture the salient dynamic characteristics with low order identified models, the physical model is evaluated for essentially nonminimal dynamics. A singular perturbation model reduction approach is then applied to obtain a minimal representation of the dynamics more suitable for control design, and yielding insight to the underlying system dynamics previously unavailable in the literature. The results demonstrate that iteratively modeling a complex system with first principles and system identification techniques gives greater confidence in the first principles model, and better understanding of the underlying physical dynamics. Although this iterative process requires more time and effort, significant insight and model improvements can be realized.

Commentary by Dr. Valentin Fuster
2004;():821-833. doi:10.1115/IMECE2004-59600.

Simplified models for predicting engine mount forces have traditionally been developed based on the assumption that for a well-balanced low-speed engine, the reciprocating dynamics can be decoupled from the three-dimensional motion of the engine block. In this paper the simplification is done systematically, using a technique previously developed by the authors to search for decoupling within a model, and to partition models in which decoupling is found. Beginning with a fully-coupled bond graph model of a balanced in-line six-cylinder engine, bonds representing negligible constraint terms are found based on aggregate power flow, and are converted to modulated sources. Separate bond graphs joined by modulating signals result. The “driving” bond graph represents the reciprocating dynamics, and the “driven” bond graph represents motion of the block on its mounts. The partitions are smaller than the original model and are simulated individually to accurately predict the dominant third-order mount forces with significant computational savings. The decoupling is found without the modeler relying on traditional assumed forms of the one-way coupled model, and can be quantitatively tracked as the system parameters and inputs change.

Commentary by Dr. Valentin Fuster
2004;():835-843. doi:10.1115/IMECE2004-59733.

Vital to the effectiveness of simulation-based design is having a system model of known quality. Previous research introduced an algorithm called AVASIM for assessing model validity systematically and quantitatively. AVASIM assess the accuracy and validity of a model based on a specific input and set of system parameters. The purpose of this paper is to present a AVASIM-based methodology that defines a range of validity of a model with respect to input and system parameter variations Two illustrative examples are presented to explore the feasibility of the proposed procedure. The first example analyzes a linearized version of a nonlinear transient vehiclehandling model. This model’s accuracy and validity are evaluated with respect to variation of two system parameters, resulting in a two-dimensional range of validity. Then a complex nonlinear hydrogen fuel cell model is linearized in order to investigate its accuracy and validity with respect to two input parameters. This again results in a two-dimensional range of validity, but with respect to input rather than system parameters. The results agree well with what is expected for the various models based on knowledge of the effects of linearization on model accuracy. The proposed algorithm for assessing model range of validity is a promising tool for determining model quality and thus potentially useful for simulation-based design.

Commentary by Dr. Valentin Fuster
2004;():845-849. doi:10.1115/IMECE2004-59789.

The authors propose the identification of stiffness using the H∞ performance criterion and a nominal state-space model of the flexible system. The parameter estimator takes the form of an Extended H∞ Filter, which is robust to uncertainties such as modeling error.

Commentary by Dr. Valentin Fuster
2004;():851-861. doi:10.1115/IMECE2004-59790.

Accurate measurements of all the state variables of a given system are often not available due to the high cost of sensors, the lack of space to mount the transducers or the hostile environment in which the sensors must be located. The purpose of this study is to design a robust sliding mode observer that is capable of accurately estimating the state variables of the system in the presence of disturbances and model uncertainties. It should be emphasized that the proposed observer design can handle state equations expressed in the general form. The performance of the nonlinear observer is assessed herein by examining its capability of predicting the rigid and flexible motions of a compliant beam that is connected to a revolute joint. The simulation results demonstrate the ability of the observer in accurately estimating the state variables of the system in the presence of structured uncertainties and under different initial conditions between the observer and the plant. Moreover, they illustrate the deterioration in the performance of the observer when subjected to unstructured uncertainties of the system. Furthermore, the nonlinear observer was successfully implemented to provide on-line estimates of the state variables for two model-based controllers. The simulation results show minimal deterioration in the closed-loop response of the system stemming from the usage of estimated rather than exact state variables in the computation of the control signals.

Commentary by Dr. Valentin Fuster
2004;():863-876. doi:10.1115/IMECE2004-59896.

The energy consumption by building heating, ventilating, and air conditioning (HVAC) systems has evoked more attention for energy efficient HVAC control and operation. Application of advanced control and operation strategies requires robust online system models. In this research, online models with parameter estimation for a building zone with variable air volume (VAV) system, which is one of the most common HVAC systems, are developed and validated using experimental data. Building zone temperature and VAV entering air flow are modeled based on physical rules and using only the measurements that are commonly available in a commercial building. Different series of validation tests were performed in a real-building test facility to examine the prediction accuracies for system outputs. Using the online system models with parameter estimation, the prediction errors for all the validation tests are less than 0.5°F for temperature outputs, and less than 50 ft3 /min for air flow outputs. The online models can be further used for local and supervisory control, as well as fault detection applications.

Commentary by Dr. Valentin Fuster
2004;():877-885. doi:10.1115/IMECE2004-60015.

The spatial structure of the dynamics of a rotating nonlinear shaft is identified by processing its finite element dynamics by the method of Proper Orthogonal Decompositions. The Proper Orthogonal modes furnish characteristic signatures for the rigid body and the whirling modes of a motion. The pattern of energy distribution over the components of a mode reveals the strength of coupling between rigid body rotations and coupled vibrations. These modes are used to derive a two-degree-of freedom reduced model for the whirling motion of the rotating shaft.

Topics: Vibration , Shapes
Commentary by Dr. Valentin Fuster
2004;():887-894. doi:10.1115/IMECE2004-60107.

Approaches to engineering design and manufacturing such as integrated design and manufacture and just in time fabrication depend on interaction with and among component supply companies that most often use very diverse technologies. Modular Distributed Modeling (MDM) is a distributed, component-based, agent methodology that is realized following a strong black box approach to modeling. An individual Design Agent (DA) is a virtual product capable of encapsulating both descriptive and model based information about the product it represents. Hierarchically recursive agents for sub-systems and/or components are linked via a communications network to form larger integrated model systems. A two dimensional bridge system structural model is used as an example to illustrate the distributed assembly of structural models from components registered as DA’s on a communications network. Modular Distributed Modeling of system dynamics performs dynamic analysis. This paper presents the methodology required to assemble dynamic structural deflection models provided by internet agents representing structural components. The methodology discussed assembles these component models into the structural dynamic model of an assembly. Using MDM method, models of complex assemblies can be built and distributed while hiding the topology and characteristics of their structural subassemblies. The automated, modular, assembly of structural dynamics models will be derived for discrete, multi-degree-of-freedom structural connections. Discrete connections are important to the assembly of components such as truss and shaft structures where the relationship between component displacements involve discrete, matching, degrees of freedom on components to be assembled. Specific examples of discrete assembly of truss bridge component models will be presented. Specific examples for distributed assembly of component models will be presented. Internet connection permitting, real-time, automated assembly of models and deflection analysis will be performed.

Commentary by Dr. Valentin Fuster
2004;():895-901. doi:10.1115/IMECE2004-60179.

A new algorithm which generates a class of input signals corresponding to which the combination of plant outputs in the output zero direction is always zero irrespective of time, is proposed. Using this result and the results in MacFarlane and Karcanias (1976) we propose two tests to find the defective rows and the defective columns of a plant transfer function matrix. Our fault detection scheme is based on the following triple: the output zero direction of the plant, the transmission zero of the plant, and the input zero direction of the plant. Based on the faulty rows and columns we identify the faulty elements of the plant transfer function matrix. A system of four interconnected water tanks with two multivariable zeros and four stable poles was found ideal for illustrating the results obtained. The algorithm proposed is used to detect changes occurring in the four-tank setup. Each input channel is analyzed one at a time. In the latter part of the work we have analyzed the 4-tank system by measuring the new steady state output of the plant which is obtained after disturbing the plant from its previous equilibrium state by changing the inputs by a small quantity from their equilibrium values. The method helps in pin-pointing the defective elements of the plant transition matrix. The key advantage of this method lies in the fact that the changes can be found without having to restart the plant.

Topics: Flaw detection
Commentary by Dr. Valentin Fuster
2004;():903-909. doi:10.1115/IMECE2004-60212.

Conventional nonlinear system identification procedures assume that the system behavior is nominally linear at a specific amplitude (usually low). The nominally linear parameters are then estimated at that particular amplitude and used to estimate the nonlinear parameters of the system. Many mechanical systems are not nominally linear over a broad frequency range for any operating amplitude. A new method for nonlinear system identification, in the absence of an input measurement, is presented that works in the opposite direction. Information about the nonlinear elements of the system is used to estimate the underlying linear parameters. Restoring force, boundary perturbation and direct parameter estimation techniques are combined to develop this approach. The approach is applied to data from an experimental tire-vehicle suspension system.

Commentary by Dr. Valentin Fuster
2004;():911-920. doi:10.1115/IMECE2004-60509.

A new tool has been developed for characterizing both the global and local dynamic behavior exhibited by peripheral circulatory waveforms. The signals representing these waveforms can be derived from a variety of different cardiovascular sensor modalities measuring changes in pressure, flow, or volume. The correlation that exists between two peripheral circulatory waveform measurements taken at different locations allows this new signal-processing algorithm to identify two compact, low order dynamic models in real-time. These models identify the distinct dynamic behavior of the circulatory paths traveled by the pulsatile waveforms using output measurements alone. This new algorithm is based on a multi-channel blind system identification technique that has been reformulated to use a Laguerre basis function series expansion. Additionally, a new deconvolution algorithm has been developed to allow estimation of the unknown common input using the identified Laguerre models, which are generally inversely unstable. This new approach has been shown to provide accurate identification of vascular hemodynamics when applied to experimental data derived from a swine.

Commentary by Dr. Valentin Fuster
2004;():921-930. doi:10.1115/IMECE2004-61431.

In this paper, a model and the associated identification procedure are proposed to precisely portray the hysteresis behavior in piezoelectric actuators. The model, presented in bond graphs, consists of basic physical elements and utilizes a Maxwell-slip structure to describe the hysteresis. By analyzing the model, the influence of the initial strain/charges on the hysteresis behavior is revealed. It is also found that if all the spring elements in the model are linear, the resultant hysteresis loop is anti-symmetric and does not match the experimental behavior. To account for this mismatch, a nonlinear spring element is included into the model. The constitutive relation of the nonlinear spring and the parameters of the basic elements in the model are identified from the experimental data using linear programming. Simulations of the identified model indicate that the model can reproduce the major as well as the minor hysteresis loops. An inverse control is further implemented to validate the accuracy of the identified model. Experiments show that the hysteresis is effectively cancelled and accurate tracking of a reference trajectory is achieved.

Commentary by Dr. Valentin Fuster
2004;():931-938. doi:10.1115/IMECE2004-59377.

The objective of this research was to use a DSP for controlling a nano-precision positioning system. The positioning system was constructed by a piezoelectric actuator, a flexural stage that composed by notch hinges and parallel springs; capacitor sensor was used instead of laser interferometer as displacement sensor. In order to increase the traveling range of the stage, a lever mechanism was used to enlarge the displacement of the piezoelectric actuator. In this study a Dmatek PICE-DSP 320C542 microcomputer developing system was used as the controller, which contained a TI TMS320C542 DSP chip. PI feedback control rule together with n-times feedforward control rule were used for the controlling of this system. Three different experiments were conducted: (1) fixed-point test; (2) continuous stepping test; and (3) ramp-tracking test. From the experiment results, it was shown that the bias Ess was within 2 nm and the standard deviation 1σ was around 30 nm, with settling time less than 0.02 sec, in the continuous stepping test. While, in the ramp-tracking test, the bias Ess was less than 1.25 nm and the standard deviation 1σ was around 35 nm.

Commentary by Dr. Valentin Fuster
2004;():939-948. doi:10.1115/IMECE2004-59473.

In this article, we present an image-based control scheme to address the problem of low operating speed in scanning tunneling microscopes (STMs). Low operating speeds are used because dynamic effects cause error in positioning the STM probe over the sample. The STM’s low operating speed limits its capability to investigate fast surface phenomena and its throughput during nanofabrication. The proposed image-based approach to achieve high speed operation exploits the extant imaging capabilities of STMs. In practice, distortions in high-speed images are used to identify and correct errors in the STM probe position. The approach is applied to an STM and experimental results are presented.

Commentary by Dr. Valentin Fuster
2004;():949-956. doi:10.1115/IMECE2004-59536.

A precision positioner using a novel concentrated-field permanent-magnet matrix is presented in this paper. This integrated multidimensional positioner is actuated by three novel planar motors, which are attached on the bottom of the positioner. It can generate all 6-DOF motions with only a single moving part. The integrated multi-dimensional positioner offers a unique combination of range and precision: a planar traveling range of 160 mm × 160 mm with a position resolution of 30 nm and position noise of 10 nm rms. The repeatability of the positioner is as good as 30 nm. The maximum velocity achieved so far is 0.5 m/s with 5-m/s2 acceleration. With direct-quadrature (DQ) decomposition theory, the positioner is modelled as a multi-input multi-output (MIMO) electromagnetic system: it has six inputs (currents) and six outputs (displacements). After the state-space model of the system is derived, multivariable controllers are designed for this high-precision positioner. To eliminate the steady-state error, discrete time integrator combined Linear Quadratic Regulation (LQR) and reduced order Linear Quadratic Gaussian (LQG) control methodologies are applied and implemented. Finally, the experimental results are presented in this paper. Several experimental results verified the utility of this positioner in precision applications, such as semiconductor manufacturing.

Topics: Modeling
Commentary by Dr. Valentin Fuster
2004;():957-964. doi:10.1115/IMECE2004-60507.

A systematic procedure for modeling and optimal control of a multivariable 6-DOF (degree-of-freedom) magnetically levitated (maglev) stage has been described in this paper. In our previous publications, we have presented the design, SISO (single-input single-output) control, and testing of the maglev stage with nanometer-precision positioning capability and several-hundred-micrometer travel range. In the present work, we extended the current model to a more rigorous LQR (linear quadratic regulation) controller for the lateral control to reduce the coupling between axes. Independent lead-lag controllers have been used for the vertical control. The system equations have been derived using the Euler angle methodology and linearized about an operating point. The performance of this multivariable control has been analyzed and compared with all the six decoupled SISO controllers. The effect of adding the integrators to eliminate the steady-state error has also been discussed and the performance of the LQR controller with different weight matrices has been compared. In this paper, we also address the issues related to the stochastic modeling of the stage to analyze the coupling between different axes and transfer function identification.

Commentary by Dr. Valentin Fuster
2004;():965-972. doi:10.1115/IMECE2004-62470.

In this paper, the behavior of nanoparticles, manipulated by an atomic force microscope nanoprobe, is investigated. Manipulation by pushing, pulling or picking nanoparticles can result in rolling, sliding, sticking, or rotation behavior. The dynamic simulation of the nanoparticle manipulation, using atomic force microscope (AFM), is performed. According to the dynamics of the system, the AFM pushing force increases to the critical value required for nanoparticle motion. Nanoparticle positioning is designed based on when the nanoparticle is stopped by the AFM in order to move on the substrate. Simulation results for gold particles on a silicon substrate showed that sliding on the substrate is dominant in nanoscales.

Commentary by Dr. Valentin Fuster
2004;():973-979. doi:10.1115/IMECE2004-59087.

This paper proposes a new trajectory-generation scheme for a high-performance anti-swing control of overhead cranes, where the trajectory-generation problem is solved as a kinematic problem. First, a new anti-swing control law is designed based on the load-swing dynamics, for which the Lyapunov stability theorem is used as a mathematical tool. Then a new trajectory-generation scheme is proposed based on the anti-swing control law and typical crane operation in practice. For g iven hoisting motions, trolley-traveling trajectory references are computed based on the concept of minimum-time control, and then anti-swing trajectories are generated based on the trajectory references through the anti-swing control law. The new trajectory-generation scheme generates a typical anti-swing trajectory in industry with high-speed load hoisting. The effectiveness of the proposed trajectory-generation scheme is shown by generating high-performance anti-swing trajectories with high hoisting speed and hoisting ratio.

Commentary by Dr. Valentin Fuster
2004;():981-990. doi:10.1115/IMECE2004-59213.

The application of haptic technology to the human-vehicle interface permits customization based on the operator’s needs, mission requirements, and surrounding environment. In mechatronic systems, haptics typically encompass force reflecting devices through the integration of sensors, actuators, and real time microprocessor control algorithms. Some transportation-based haptic interface technology examples include ground vehicles, electric wheelchairs, remote-controlled drones, aircraft, and mobile robots. Drive-by-wire vehicles and tele-operated robots can benefit from haptic concepts through the provision of tunable force feedback using servo-motors to upgrade traditional systems such as hydraulic power steering and non-force feedback joysticks. In addition, haptics can potentially improve the operator’s performance and overall system safety. In this paper, driver interface feedback has been studied on a steer-by-wire haptic interface platform integrated with a virtual reality driving environment. Operator performance over a prescribed series of driving maneuvers and feedback settings has been investigated through real time data logging and post-test questionnaires. Finally, external observations have been made on individual driver behavior when subjected to varying steering feedback configurations.

Commentary by Dr. Valentin Fuster
2004;():991-1001. doi:10.1115/IMECE2004-59239.

In this paper an effective approach for kinematic and dynamic modeling of high mobility wheeled mobile robots (WMR) has been presented. As an example of these robots, the method has been applied on CEDRA rescue robot which is a complex, multibody mechanism. The model is derived for 6-DOF motions enabling movement in x, y, z directions, as well as pitch, roll and yaw rotations. Forward kinematics equations are derived using Denavit-Hartenberg method and the wheels Jacobian matrices. Moreover the inverse kinematics of the robot is obtained and solved for the wheel velocities and steering commands in terms of desired velocity, heading and measured link angles. Finally dynamical analysis of the rover has been thoroughly studied. Due to the complexity of this multi-body system especially on rough terrain, Kane’s method of dynamics has been used to model this problem. The approach has been developed in such a way that it can easily be extended to other mechanisms and rovers.

Commentary by Dr. Valentin Fuster
2004;():1003-1010. doi:10.1115/IMECE2004-59351.

Programmable mechanical compliance (PMC) in actuation is desirable for human interaction tasks and important for producing biomimetic motion. In this paper, the equilibrium point (EP) hypothesis is proposed and implemented as a strategy for controlling programmable compliance. A two-DOF planar manipulator activated by McKibben actuators was constructed for the purpose of demonstrating the application of the equilibrium point hypothesis on a robotic platform. The equilibrium angle and stiffness of each of the joints on the manipulator can be independently programmed. The results presented herein show stable behavior for free motion, interaction and transition tasks using the EP hypothesis implemented with a linear PI control strategy.

Commentary by Dr. Valentin Fuster
2004;():1011-1018. doi:10.1115/IMECE2004-59353.

A high-quality haptic interface is typically characterized by low apparent inertia and damping, high structural stiffness, minimal backlash and absence of mechanical singularities in the workspace. In addition to these specifications, exoskeleton haptic interface design involves consideration of additional parameters and constraints including space and weight limitations, workspace requirements and the kinematic constraints placed on the device by the human arm. In this context, we present the design of a five degree-of-freedom haptic arm exoskeleton for training and rehabilitation in virtual environments. The design of the device, including actuator and sensor selection, is discussed. Limitations of the device that result from the above selections are also presented. The device is capable of providing kinesthetic feedback to the joints of the lower arm and wrist of the operator, and will be used in future work for robot-assisted rehabilitation and training.

Topics: Design , Haptics
Commentary by Dr. Valentin Fuster
2004;():1019-1028. doi:10.1115/IMECE2004-59428.

In this paper, we examine and evaluate candidate articulated leg-wheel subsystem designs for use in vehicle systems with enhanced uneven-terrain locomotion capabilities. The leg-wheel subsystem designs under consideration consist of disk wheels attached to the chassis through an articulated linkage containing multiple lower-pair joints. Our emphasis is on creating a design that permits the greatest motion flexibility between the chassis and wheel while maintaining the smallest degree-of-freedom (d.o.f.) within the articulated chain. In particular, we focus our attention on achieving two goals: (i) obtaining adequate ground clearance by designing the desired/feasible motions of the wheel axle, relative to the chassis, using methods from kinematic synthesis; and (ii) reducing overall actuation requirements by a judicious mix of structural equilibration design and spring assist. We examine this process in the context of two candidate designs — a coupled-serial-chain configuration and four-bar-configuration — for the articulated-leg-wheel subsystem. The performance of planar variants of these designs, operating in the sagittal plane, is evaluated and representative results are presented to highlight the process.

Topics: Design , Wheels
Commentary by Dr. Valentin Fuster
2004;():1029-1036. doi:10.1115/IMECE2004-59593.

This paper discusses the use of multiple vision sensors and a proximity sensor to obtain three-dimensional occupancy profile of robotic workspace, identify key features, and obtain a 3-D model of the objects in the work space. The present research makes use of three identical vision sensors. Two of these sensors are mounted on a stereo rig on the sidewall of the robotic workcell. The third vision sensor is located above the workcell. The vision sensors on the stereo rig provide information about three-dimensional position of any point in the robotic workspace. The camera to robot calibration for these vision sensors in stereo configuration has been obtained with the help of a three-layered feedforward neural network. Squared Sum of Difference (SSD) algorithm has been used to obtain the stereo matching. Similarly, camera to robot transformation for the camera located above the work cell has been obtained with the help of a three-layered feedforward neural network. Three-dimensional positional information from vision sensors on stereo rig and two-dimensional positional information from a camera located above the workcell and a proximity sensor mounted on the robot wrist have been fused with the help of Bayesian technique to obtain more accurate positional information about locations in workspace.

Topics: Sensors , Robotics
Commentary by Dr. Valentin Fuster
2004;():1037-1041. doi:10.1115/IMECE2004-59621.

Medical applications are among the most fascinating areas of microrobotics. For long, scientists have dreamed of miniature smart devices that can travel inside the human body and carry out a host of complex operations such as minimally invasive surgery (MIS), highly localized drug delivery, and screening for diseases that are in their very early stages. Still a distant dream, significant progress in micro and nanotechnology brings us closer to materializing it. For such a miniature device to be injected into the body, it has to be 800 μm or smaller in diameter. Miniature, safe and energy efficient propulsion systems hold the key to maturing this technology but they pose significant challenges. Scaling the macroscale natation mechanisms to micro/nano length scales is unfeasible. It has been estimated that a vibrating-fin driven swimming robot shorter than 6 mm can not overcome the viscous drag forces in water. In this paper, the authors propose a new type of propulsion inspired by the motility mechanism of bacteria with peritrichous flagellation, such as Escherichia coli, Salmonella typhimurium and Serratia marcescens. The perfomance of the propulsive mechanism is estimated by modeling the dynamics of the motion. The motion of the moving organelle is simulated and key parameters such as velocity, distribution of force and power requirments for different configurations of the tail are determined theoretically. In order to validate the theoretical result, a scaled up model of the swimming robot is fabricated and characterized in silicone oil using the Buckingham PI theorem for scaling. The results are compared with the theoretically computed values. These robots are intended to swim in stagnation/low velocity biofluid and reach currently inaccessible areas of the human body for disease inspection and possibly treatment. Potential target regions to use these robots include eyeball cavity, cerebrospinal fluid and the urinary system.

Topics: Propulsion
Commentary by Dr. Valentin Fuster
2004;():1043-1050. doi:10.1115/IMECE2004-59678.

The control problem of the spatial tentacle manipulator is presented. In order to avoid the difficulties generated by the complexity of the nonlinear integral - differential model, the control problem is based by the artificial potential method. It is shown that the control of a tentacle robot to a desired position it is possible if the artificial potential is a potential functional whose point of minimum is attractor of this dissipative controlled system. Then, the method is used for constrained motion in an environment with obstacles. Numerical simulations for spatial and planar tentacle models are presented in order to illustrate the efficiency of the method.

Topics: Manipulators
Commentary by Dr. Valentin Fuster
2004;():1051-1058. doi:10.1115/IMECE2004-59743.

Digital Clay is a proposed novel three-dimensional computer input and output device for surface shape and haptic effects. The device consists of an array of fluidic actuators under the control of valves connected to two pressure reservoirs in a manner ultimately suitable to an implementation in MEMS technology. The challenges to build this device lie in both the kinematical structure design and the control architecture. Though it is proposed to ultimately build the actuators and control valves using MEMS technology, conventional methods are used at the current prototype stage. In this paper, several designs of the practical kinematical structure will be discussed and the proposed control architecture for the Digital Clay will be introduced.

Topics: Design
Commentary by Dr. Valentin Fuster
2004;():1059-1064. doi:10.1115/IMECE2004-59792.

The paper presents the kinematic synthesis of a symmetric parallel platform supported by three RRS serial chains. The dimensional synthesis of this three degree-of-freedom system is obtained using design equations for each of three RRS chains obtained by requiring that they reach a specified set of task positions. The result is 10 polynomial equations in 10 unknowns, which is solved using polynomial homotopy continuation. An example is provided in which the direction of the first revolute joint (2 parameters) and the z component of the base and platform are specified as well as the two task positions. The system of polynomials has a total degree of 4096 which means that in theory it can have as many solutions. Our example has 70 real solutions that define 70 different symmetric platforms that can reach the specified positions.

Topics: Design
Commentary by Dr. Valentin Fuster
2004;():1065-1073. doi:10.1115/IMECE2004-59805.

The “virtual wall” is the most common building block used in constructing haptic virtual environments. A virtual wall is typically based on a simple spring model, with unilateral constraints that allow the user to make and break contact with a surface. There are a number of factors (sample-and-hold, device dynamics, sensor quantization, etc.) that cause virtual walls to demonstrate energetic behavior, destroying the illusion of reality. Passive objects are incapable of generating energy, so in this paper, we find an explicit upper bound on virtual wall stiffness that is a sufficient condition for virtual wall passivity. We consider a haptic display that can be modeled as a mass with Coulomb and viscous friction, being acted upon by two external forces — an actuator and a human user. The system is equipped with only one sensor — an optical encoder measuring the position of the mass. We explicitly consider the effects of position resolution, which has not been done in previous work. We make no assumptions about the human user, and we consider arbitrary constant sampling rates. The main result of our analysis is a sufficient condition for passivity that relies on the Coulomb friction in the haptic device.

Commentary by Dr. Valentin Fuster
2004;():1075-1082. doi:10.1115/IMECE2004-59830.

In this paper, the motion of an elastically suspended rigid body unilaterally constrained by frictional contact at multiple locations is studied. In this problem, each individual contact may or may not constrain the body’s motion. The set of actual active constraints is determined by: 1) the commanded motion of the body’s compliant support, 2) the coefficient of friction at each contact point, 3) the number and geometry of all potentially constraining surfaces, and 4) the elastic properties of the support. Here, the investigated problem is restricted to quasistatic motion and the interaction is characterized by Coulomb friction. We show that, for a passive compliant system, if the coefficient of friction at each contact is upper bounded, the set of active constraints is unique. A procedure to determine both the set of active constraints and the motion of the constrained body is provided.

Commentary by Dr. Valentin Fuster
2004;():1083-1089. doi:10.1115/IMECE2004-59874.

The design and evaluation of a new controller for a multi-finger tactual (kinesthetic and cutaneous) display, the TACTUATOR, is discussed. A crucial performance requirement is that the relative amplitude of spectral components be preserved in terms of perceived intensity as judged by human users. In this article, we present a two degree-of-freedom controller consisting of a feedback controller and a pre-filter, and its digital implementation. The overall system was evaluated with frequency-response function measurements and with human psychophysical experiments. The measurement results confirm that the steady-state frequency response closely follows the design specifications. The psychophysical results indicate a deviation in the model of the human detection threshold curve at frequencies below 30 Hz. Future work will compensate for this deviation by reshaping the pre-filter. Our work demonstrates the validity of designing controllers that take into account not only the electromechanical properties of the hardware, but the sensory characteristics of the human user.

Commentary by Dr. Valentin Fuster
2004;():1091-1098. doi:10.1115/IMECE2004-60049.

Providing the user with high-fidelity force feedback has persistently challenged the field of telerobotics. This paper presents a new approach for achieving stable, high-gain force reflection via cancellation of the master mechanism’s induced motion. In a classic position-force controller, high force-feedback levels drive the system’s internal master-slave loop unstable during contact with the remote environment. Lowering the force-feedback gain ensures stability but diminishes the haptic cues available to the user, masking contacts and preventing hard objects from feeling appropriately stiff. The proposed cancellation approach permits high levels of force feedback by attenuating only the controller’s internal loop. Using a model of the master mechanism’s response to applied force feedback, an estimate of induced high-frequency movement is substracted from the master’s measured position to approximate the user’s intended path for the slave. The cancellation technique is described, modeled, and validated herein, including testing on a one-degree-of-freedom telerobotic system. It is shown to improve the feel of the system, tripling the testbed’s achievable force-feedback gain without compromising stability.

Topics: Force , Motion
Commentary by Dr. Valentin Fuster
2004;():1099-1107. doi:10.1115/IMECE2004-60058.

The task of complete reconfiguration of a rolling sphere is recast as a problem in planar geometry. Based on our specific choice of coordinate system, two control points are defined that equivalently represent the configuration of the sphere. Under the applied control inputs, the trajectories of these control points are shown to form an evolute-involute pair. The objective of complete reconfiguration is then captured within integral conditions over the path of the control points. A class of solution to the geometric problem is derived and a three step algorithm for complete reconfiguration is proposed. The solution is computationally in-expensive and is confirmed by simulation results.

Topics: Geometry
Commentary by Dr. Valentin Fuster
2004;():1109-1121. doi:10.1115/IMECE2004-60060.

This paper describes the mechanical and electrical design of a compact high fidelity desktop haptic interface that provides three-degree-of-freedom point-force interaction through a handheld pen-like stylus. The complete haptic device combines a spatial linkage, actuation, power amplification, and control electronics in a standalone package with a footprint similar to that of a notebook computer (33cm × 25cm × 10cm). The procedure used to design the statically balanced spatial linkage is explained and both an inexpensive lightweight plastic version and a high stiffness, high strength, aluminum and stainless steel version are presented. The theory and implementation of sinusoidal encoder interpolation and sinusoidal servo-motor commutation used to achieve high-fidelity haptic simulation is covered for two versions of electronic control hardware: custom hardware based on a digital signal processor (DSP) and an off-the-shelf design based on an embedded PC.

Topics: Linkages , Design , Haptics
Commentary by Dr. Valentin Fuster
2004;():1123-1130. doi:10.1115/IMECE2004-60064.

A networked control system (NCS) is a control architecture where sensors, actuators and controllers are distributed and interconnected. It is advantageous in terms of interoperability, expandability, installation, volume of wiring, maintenance, and cost-effectiveness. Many distributed network systems of various topologies and network type have been developed, but NCS systems tend to suffer from such issues as nondeterminism, long network delays, large overheads and unfairness. This paper presents the ring-based protocol, called the ExoNet, and its network architecture which are built to achieve better performance as a distributed networked system. A Cypress transceiver CY7C924ADX is applied to the network as a communication unit. The protocol is based on the transceiver and developed to achieve fast communication and allowable latency for controls with high control loop frequency. Compared with other standard network types such as Ethernet, ControlNet or DeviceNet, the network is characterized by its ring-based architecture, simple message and packet formats, one-shot distribution of control data and collection of sensor data, multi-node transmission, echo of a message, and other features. The network also guarantees determinism, collision-free transmission, relatively small overhead, fairness between nodes and flexibility in configuration. Its analysis and comparison with these network types are also provided and its application on the Berkeley Lower-Extremity Exoskeleton (BLEEX) is described.

Topics: Control systems
Commentary by Dr. Valentin Fuster
2004;():1131-1138. doi:10.1115/IMECE2004-60137.

In this paper, we present a frequency-based trajectory planning approach that considers the variable dynamic bandwidth of any multibody system having different dynamic subsystems. This approach provides us with important motion planning methodology for a variety of robotic systems including land-based mobile robots, space robots, and underwater robots, where the vehicles have much slower response as compared to the manipulators. The proposed method has been improvised for an Autonomous Underwater Vehicle-Manipulator System (UVMS) which is a heterogeneous dynamic system having vehicle’s natural frequency much lower than that of the manipulator. This motion-planning algorithm not only considers the variability in dynamic bandwidth of such a complex system but also generates kinematically admissible as well as dynamically feasible reference trajectories. Additionally, the proposed algorithm exploits the inherent kinematic redundancy of the system and provides reference trajectories that accommodate several other important criteria such as thruster/actuator fault and saturation; it also minimizes hydrodynamic drag on the UVMS. Here, we have mainly compared the performance of two frequency-based decomposition approaches, namely: Partial Decomposition and Total Decomposition. The effectiveness of the proposed algorithm is verified with extensive computer simulations and the results are found quite promising.

Commentary by Dr. Valentin Fuster
2004;():1139-1146. doi:10.1115/IMECE2004-60190.

Mobile inverted pendulum robots consist of an elongated pendulum body with two motors mounted on it for driving the wheels. The velocity and position control of such a vehicle is challenging because of the coupling of the pendulum’s pitch angle from the vertical and the Cartesian motion of the vehicle. On using a nonlinear transform to effect partial feedback linearization, the system dynamics transforms to two subsystems: a linear system with the vehicle’s pitch and in-plane orientation, and a nonlinear system of internal dynamics. In this paper, the problem of utilizing such a partial feedback linearization for optimal trajectory planning of such vehicles is considered. Due to the resulting linear subsystem with vehicle pitch and in-plane orientation, these configuration variables can be controlled by a linear controller (C) using a nonlinear feedback. The planning approach presented in this paper considers the time-constant (τ) of the linear controller C explicitly. A band-limited Sinc-function interpolation is used to plan the variables in the method of collocation. This ensures that high frequency signal content which cannot be handled by the controller C, is absent in the planned trajectory, making the plan better implementable by the controller. The planned trajectory takes the vehicle from point to point, while keeping the vehicle pitch bounded and avoiding obstacles. The optimality condition considered in the algorithm is the length of the path. The optimization problem posed after collocation, is then solved using a standard Sequential Quadratic Programming (SQP) solver. Simulation results show that the tracking controller C is able to follow the planned trajectory when no initial planar Cartesian error is present.

Commentary by Dr. Valentin Fuster
2004;():1147-1153. doi:10.1115/IMECE2004-60206.

Ground autonomous mini-mobile robots have important potential applications, such as reconnaissance, patrol, planetary exploration and military applications. To accomplish tasks on rough-terrain, control and planning methods must consider the physical characteristics of the vehicle and of its environment. Failure to understand these characteristics could lead to vehicle endangerment and mission failure. This paper describes recent and current work at Mobile Robotics Laboratory of the Politecnico of Bari in the area of rough terrain mobility and traversability of autonomous vehicles. A cylindrical shaped mobile robot is presented and its rolling motion on rough terrain is studied from both theoretical and experimental prospect. A comprehensive vehicle dynamic model is proposed based on well-established physical models of mobile robot-terrain interaction. The model is experimentally validated and it allows employing the vehicle as a tactile sensor for terrain characterization and identification. Innovative vision-based-methods are also introduced for estimating relevant kinematic parameters of the vehicle motion. It is shown that the dynamic model can describe efficiently the vehicle behavior and could enhance its mobility on rough-terrain through integration with control and planning algorithms.

Commentary by Dr. Valentin Fuster
2004;():1155-1162. doi:10.1115/IMECE2004-60217.

In this paper, the impact dynamics and its effects on the contact event of a five-link biped walking on level ground are studied. The dynamic equations of both single and double impact of the biped are derived using the Newtonian approach. We improved the conventional five-link biped kinematic model such that, for the first time, the explicit solutions of the external impulses and post impact angular velocities are presented in detailed but compact form. These solutions are used conveniently to evaluate the conditions that warrant each individual type of impact. A parametric study is performed to correlate the type of impact with certain gait parameters and the results are presented in graphical form in the parameter space. Effects of variations in the parameters on the vertical impulse and post impact velocity of the trailing tip are further investigated. The ratio of the vertical and horizontal impulses is also examined, which shows the likelihood of slippage of the foot (feet). This work reveals important physical insights into the mechanisms of biped impact and facilitates motion planning and motion regulation of biped locomotion with prescribed contact phases.

Commentary by Dr. Valentin Fuster
2004;():1163-1170. doi:10.1115/IMECE2004-60411.

A substructuring approach to derive dynamic models for closed-loop mechanisms is applied to model a flexible-link planar parallel platform with Lead Zirconate Titanate (PZT) transducers. The Lagrangian Finite Element (FE) formulation is used to model flexible linkages, in which translational and rotary degrees of freedom exist. Craig-Bampton mode sets are extracted from these FE models and then used to assemble the dynamic model of the planar parallel platform through the application of Lagrange’s equation and the Lagrange multiplier method. Electromechanical coupling models of surface-bonded PZT transducers with the host flexible linkages are introduced to the reduced order dynamic models of flexible linkages. The assembled system dynamic model with moderate model order can represent essential system dynamic behavior and maintain kinematic relationships of the planar parallel platform. A Proportional, Integral, and Derivative (PID) control law is used as the motion control law. Strain rate feedback (SRF) active vibration control is selected as the vibration control law. Motion control simulation results with active vibration control and simulation results without active vibration control are compared. The comparison shows the effectiveness of active vibration control.

Commentary by Dr. Valentin Fuster
2004;():1171-1177. doi:10.1115/IMECE2004-60450.

This paper presents a novel tactile navigation system for the blind. The device is portable and cost effective, and will allow visually impaired individuals to navigate through familiar and non-familiar environments without relying on the assistance of a guide. The “Tactile Handle” consists of an array of vibro-tactile actuators positioned to match the finger phalanxes, proximity sensors, and an embedded micro-controller. The handle processes sensor signals and outputs information to the user through variable and synchronized vibrations, which will enhance the sense of orientation and distance for the user. The prototype has an ergonomic design, is lightweight, compact, and adjustable to different hand sizes. This paper describes the concept of its use as a navigation system for the visually impaired and the preliminary results for the tactile perception of 30 sighted users of different genders and hand sizes.

Topics: Feedback , Navigation
Commentary by Dr. Valentin Fuster
2004;():1179-1186. doi:10.1115/IMECE2004-60525.

An efficient real time path planning method for groups of mobile robots in dynamic environments is introduced. Harmonic potential functions are utilized along with the panel method known in fluid mechanics. First, a complement to the traditional panel method is introduced to generate a more effective harmonic potential field for obstacle avoidance in dynamically changing environments. Second, a group of mobile robots working in an environment containing stationary and moving obstacles is considered. Each robot is assigned to move from its current position to a goal position. The group is not forced to maintain a formation during the motion. Every robot considers the other robots of the group as moving obstacles and hence the physical dimensions of the robots are also taken into account. The path of each robot is planned based on the changing position of the other robots and the position of stationary and moving obstacles. Finally, the effectiveness of the scheme is shown by modeling groups of an arbitrary number of mobile robots and the theory is validated by several computer simulations and hardware experiments.

Topics: Mobile robots
Commentary by Dr. Valentin Fuster
2004;():1187-1194. doi:10.1115/IMECE2004-61132.

In this article kinematic calibration of the central linkage of a wire-actuated parallel robot, which has a parallelogram mechanism in its structure, is discussed. It is shown that the dependency between the parallelogram joint angles affect the conditioning of the identification Jacobian matrix. Through a sensitivity analysis, it is presented that some of the parallelogram link lengths are not identifiable. Two nonlinear calibration approaches, Gauss-Newton and Levenberg-Marquardt, have also been explained and their difference especially when the identification Jacobian matrix is ill-conditioned, are pointed out.

Commentary by Dr. Valentin Fuster