State Estimation of a Muscle-Driven Linkage

This paper addresses the problem of state estimation of a two-joint six-muscle linkage. It is assumed that noisy measurements of the joint angles are available. It is shown that the joint's velocities, tendon-muscle lengths and rates of tendon-muscle lengths can be estimated using a high-gain observer (HGO), sliding mode observer (SMO), and extended Kalman filter (EKF). The estimated states are passed to the controller of the system, which is a backstepping controller. The results verify the effectiveness of the observers for the state estimation of the two-joint six-muscle model. The standard deviations of the estimation errors are very small, on the order of 10-4-10-2 rad for joint angles, 10-3-10-1 rad/s for joint velocities, 10-4-10-3 m for muscle-tendon length, and 10-5-10-3 m/s for muscle-tendon length rates. In general the HGO and EKF performed better than the SMO for this system and is less sensitive to measurement noise.