Otolith-Canal Optimal Interaction for Estimating Body Angular Velocity and Orientation Based on Two Kalman Filters

The human vestibular system, consisting of the otolith and semicircular canals, provides valuable information about body motion and orientation in terms of angular velocity and translational acceleration. This article discusses the dynamic characteristics of these organs and the generated signals and proposes approaches to combine these signals to infer for the body angular velocity. This is the first step towards determining the body orientation and thus the inertial acceleration. The first presented approach evolves around a complementary filter that capitalizes on the low- and high-frequency characteristics of the signals. The second approach resembles two versions of Kalman filter that allow for explicitly estimating the drift of the semicircular canal signal. The difference between the two versions is the number of states employed: the first relies on two states while the second employs innovatively four states selected to minimize the estimation error of orientation, velocity and drift.