Model Predictive Control for Parallel Vision/Force Control of the Collaborative Robot

In this paper, a novel parallel vision/force model predictive control methodology is developed to deal with the compliant interaction control of the collaborative robot. The visual information and the force sensing are coupled in the image feature space in parallel. In order to avoid the overshoot of the interaction force control for most cases and consider several constraints of the image-based visual servoing system, the model predictive control (MPC) is proposed. With this controller, the desired interaction force command is set to be a constraint, and the overshoot of the force control can be eliminated. With the help of the dynamic IBVS model of the robot, the constraints of the actuator torque, joint angle, and visibility can be satisfied simultaneously. The numerical simulation results for the 2 degrees of freedom (DOF) collaborative robot with an eye-to-hand camera demonstrate the efficacy of the proposed controller.