Model Predictive Force Control for Robots in compliant Environments with guaranteed Maximum Force

In this paper, we propose a model predictive force controller, which ensures that the force the object in an unknown environment exerts on the robot is limited by a user-defined value. The environment is assumed compliant and during movement, the internal model of the environment is constantly updated. This update is performed using position and force information. The information of past measurements is stored in a memory function in the form of additional constraints for the MPC. The proposed controller is tested in simulation where it has to overcome both convex and nonconvex obstacles in an unknown environment.