Prescribed-Time, Decentralized and Delay-Adaptive Control Strategies for Robot Manipulators: Design and Experiments

In this manuscript, we formulate and experimentally verify four state-of-the-art controlstrategies on Baxter, a 7-DOF redundant robot manipulator. The control strategies examined in this manuscript are the subject of active research in the field of non-linear control, and have the potential to significantly improve the performance of robot manipulators when they operate in unstructured environments. The first control strategy we investigate in this manuscript is model-free decentralized-adaptive control. The purpose of this control strategy is to achieve consistent performance across a wide range of joint configurations and end-effector inertias, while having a similar computational efficiency as PID approaches. The second control strategy we investigate in this manuscript is delay-adaptive control. The purpose of this control strategy is to simultaneously estimate and compensate for an unknown long actuator delay. The third control strategy we investigate in this manuscript is prescribed-time control. A key feature of this control strategy is that the settling time is explicitly assigned by the control designer to a value desired, or “prescribed” by the user, and that the settling time is independent of the initial conditions and of the reference signal. The fourth control strategy we investigate in this manuscript is the prescribed-time safety filter. This formation yields a filter that is capable of avoiding multiple obstacles in a minimally invasive manner with bounded joint torques, while simultaneously allowing a nominal controller to converge to positions located on the boundary of the safe set by the end of a fixed-duration task. Through the formulation and experimental verification of each control strategy we present in this manuscript, we demonstrate that our proposed methods perform well in both theory and in practice