Energy-aware adaptive impedance control using offline task-based optimization

Abstract

Robotic tasks involving interaction with an environment are commonly addressed by means of impedance control. Under the framework of energy-aware robotics, and motivated by evidence that many tasks benefit from variable impedance, we leverage a geometric formulation of an impedance controlled manipulator with time-varying impedance, to design an optimisation problem able to produce an open-loop, task-based control action. A feedback controller accounts for model variations and disturbances from the nominal task and completes the control strategy. Beyond achieving passivity by means of energy tanks, a novel energy budgeting protocol ensures safety along task execution. Simulations on a 5-DoF robot executing a peg-in-hole task validate the approach.