Observer-Based Adaptive Robust Force Control of a Robotic Manipulator

Maintaining precise interaction force in uncertain environments characterized by unknown and varying stiffness or location is significantly challenging for robotic manipulators. Existing approaches widely employ a two-level control structure in which the high level generates the command motion of the low level according to the force tracking error. However, the low-level motion tracking error is generally ignored completely. Recognizing this limitation, this paper first formulates the low-level motion tracking error as an unknown input disturbance, based on which a dynamic interaction model capturing both structured and unstructured uncertainties is developed. With the developed interaction model, an observer-based adaptive robust force controller is proposed to achieve accurate and robust force modulation for a robotic manipulator using a force/torque(F/T) sensor. Specifically, the structured uncertainties, such as the unknown and varying environmental location or stiffness, are estimated and attenuated by the synthesized adaptive law. Meanwhile, the majority of the unstructured uncertainties including modeling errors and motion tracking errors arising from the low-level controller are estimated by the extended state observer(ESO) and subsequently compensated via feedforward. Furthermore, the remaining unstructured uncertainties are effectively attenuated by the robust feedback control law. Alongside the theoretical stability analysis, comparative experiments on a robotic manipulator are carried out in four cases, and the experimental results show the significant superiorities of the propoed approach over existing methods in terms of accuracy and robustness of the interaction force control.