Extended Kalman Filtering for Robot Joint Angle Estimation Using MEMS Inertial Sensors

The possibility of utilizing low-cost MEMS accelerometers and gyroscopes for accurate joint position estimation of the robot manipulator is investigated in this paper. Cascade Kalman filtering formulation is derived from the robot forward kinematics and the stochastic models of the joint motion sensors. We validate the accuracy of the proposed algorithm via experimentation. We also discuss the effect of the nonlinearity in the kinematic model on two approximation methods - the first-order linearization and the unscented transform.