Constrained-Differential-Kinematics-Decomposition-Based NMPC for Online Manipulator Control with Low Computational Costs

Flexibility combined with the ability to consider external constraints comprises main advantages of nonlinear model predictive control (NMPC). Applied as a motion controller, NMPC enables applications in varying and disturbed environments, but requires time-consuming computations. Hence, given full multi-DOF robot model, delay-free execution providing short horizons at appropriate prediction for accurate motions is not applicable common use. This contribution introduces an approach that analyzes decomposes differential kinematics similar inverse method assign Cartesian boundary conditions specific systems equations during building, reducing online computational costs. The resulting fully constrained realizes translational obstacle avoidance trajectory tracking using reduced considering both joint coupled Jacobian transposed controller performing end-effector’s orientation correction. Apart from safe distance obstacles, presented does lead any limitations reachable workspace, all degrees freedom (DOFs) are used. simulative evaluation Gazebo Stäubli TX2-90 commanded ROS on standard computer emphasizes significantly lower costs, accuracy analysis, extended adaptability avoidance, additional flexibility. An interpretation new concept discussed further use extensions.