A Nonlinear PD Controller for Force and Contact Transient Control

his article describes a nonlinear proportional and derivative T ( N P D ) controller and its use for both non-contact transient force control (NCTFC) (zero velocity when contact occurs) and contact transient force control (CTFC) (non-zero velocity when contact occurs). The key advantages of NPD control are its high disturbance rejection and robustness to time delay. We present a gain design method for NCTFC. Simulations and experiments on the Sarcos Dextrous Arm show that PD control becomes unstable for CTFC while NPD control is stable and reaches the steady state quickly. Experiments with human subjects on NCTFC and CTFC are also presented. Despite performance limitations, PD control is most widely used for robot position control because of its simplicity. More advanced position controllers often incorporate PD control, such as computed torque or resolved acceleration control [7]. For robot force control, most controllers are simple PD or PID [ 1,2, 13,151 as well. With a rigid environment, PD based force control often induces instability and large oscillations because the loop gain is too high [ l ] and there is not enough damping. The performance is even worse than that of open loop force control, which is always stable, but open loop force control has a low bandwidth and is sensitive to force disturbances. In any real force control task, a robot needs to frequently make and break contact with the environment. The robot must successfully manage two states-free motion and contact-and state transitions. The state transition adds discontinuity and nonzero velocity. These transitions may be event-driven or simply caused by large disturbances. All these problems make the transient controller design a challenging issue. A number of force control strategies have been proposed: impedance control [3], stiffness control [lo], passive and active damping controls [6, 9, 141, and command preshaping control