Passive arbitration in adaptive shared control of robots with variable force and stiffness scaling

In mixed-initiative shared control, the final control command to the robot is a weighted sum of the commands from two or more agents (human operators or automatic control systems). In force controlled robots, scaling of forces without power-consistent scaling of velocities leads to loss of passivity of the overall system. In this work, we first pose the problem statement related to position drift, while using a state-of-the-art, passivity ensuring method for scaling of forces. We then formulate adaptive mixed-initiative shared control as an adaptive stiffness control approach. We ensure passivity of the adaptive stiffness controller with a novel, model-independent method. The salient features, benefits and limitations of the approach are emphasized through analyses, simulations and hardware experiments. The proposed approach is finally validated with a practical shared control task.