Towards the development of a humanoid arm by minimizing interaction forces through minimum impedance control

We present the results from our work in the development of a robotic arm with minimal impedance. The development of such an arm is useful for gentle exploration of unknown objects in unstructured environments. Similar to a human, the robotic arm should minimize the contact forces in the event of unanticipated contact with unknown objects in the absence of visual feedback. To accomplish this, our strategy is to develop a good model of the robotic system so that we can use low gains which in turn will lead to low impedance and hence low contact forces in manipulation tasks in unstructured environments. The paper demonstrates how good modeling and feedforward compensation can result in low interaction forces without any external force sensing. We present experimental results to demonstrate the validity of our model and approach.