Energy-aware adaptive interaction control using offline task-based optimization in an impedance control framework

Impedance control is a common form of robot control where interaction is important. A key question posed in this paper is how the impedance can be chosen in a structured manner given the robot's task and dynamics. We exploit impedance control ideas and make them optimal in time resulting in an open-loop control action which mimics time-varying Cartesian impedance control. We utilize energy tanks to recover passivity, in view of safety. In addition, we propose an iterative feed-forward adaptation law to account for model variations from the nominal plant using open-loop control. Simulation studies on a 5-DoF robot show efficacy of the methods by successfully performing an energy-aware and task-based peg-in-hole task using open-loop control with minimal feedback.