Optimal Tracking Controller Design for Invariant Dynamics Direct-Drive Arms

This paper presents an optimal solution to the problem of tracking controller design for a category of direct-drive robot arms, mechanically constructed to have invariant and decoupled joint actuator dynamics. For good tracking behavior, the controller uses future reference positions of a joint to anticipate the changes in reference velocity. An explicit acceleration feedforward term is avoided improving the power to noise ratio of the control signal. For good regulation behavior, the controller uses position and velocity feedback. An integral of error term is also avoided, reducing the probability of the occurence of limit cycle oscillations caused by saturation of the actuator torque rating.