Learning control for impedance controlled manipulator

The authors propose a model-based learning control method for impedance controlled manipulators interacting with environments to perform accurate tracking of desired force and position trajectories. A Newton-like method is used to obtain the learning control law, which includes tracking errors and estimated parameters of environments such as viscous, elastic and friction coefficients. By choosing these estimated parameters appropriately, the method is applicable to both viscoelastic and rigid environments with friction. Sufficient conditions on estimated parameters are given for the convergence of executed trajectories to the desired trajectories. The results of simulation experiments verify the effectiveness of the developed control scheme