Human-Robot Collaboration using Variable Admittance Control and Human Intention Prediction

Due to the difficulty of modeling human limb, it is very challenging to design the controller for human-robot collaboration. In this paper, we present a novel controller combining the variable admittance control and assistant control. In particular, the reinforcement learning is used to obtain the optimal damping value of the admittance controller by minimizing the reward function. In addition, we use the long short-term memory networks (LSTMs) to predict human intention based on the human limb dynamics and then an assistant controller is proposed to help human complete collaboration tasks. We validate the performance of our prediction algorithm and controller on a 7 d.o.f Franka Emika robot equipped with joint torque sensors. The proposed controller can both achieve minimum-jerk trajectory and low-effort cost.