Stewart Platform Manipulator: State Estimation Using Inertia Sensors and Unscented Kalman Filter

This manuscript presents the estimation of both position and velocity of Stewart manipulator using leg length measurements and MEMS inertial sensors. The estimation by using Unscented Kalman Filter (UKF) is based on the combination of different sensors. UKF is used as a nonlinear state estimator to the Stewart platform which is modeled as a stochastic differential equation due to measurement noise. The experimental results of the UKF are verified on the Stewart platform DELTALAB EX800 using LABVIEW real time software. The desired trajectories are compared with the estimated states (position and orientation) obtained using UKF. Moreover, the estimated leg lengths are compared with the real measured output from the potentiometer sensors of the six legs of the parallel manipulator. The experimental results show that the estimation error is bounded with small bound depending on the covariance matrices. This proves the effectiveness of the proposed Unscented Kalman Filter (UKF) as a nonlinear estimator with integration between inertial sensors and leg potentiometers.