This project focuses on general framework for semi-autonomous teleoperation of robot manipulators via model predictive control (MPC). The hand pose of the human operator provides the reference for the end effector, and the robot motion is continuously replanned in real time, satisfying several constraints (including limits on joint accelerations, velocities and positions, avoidance of near-singular configurations, and obstacle avoidance for the whole robot frame).
The motion capture is done with a custom wireless inertial setup build in-house. The robot used in the project is the Universal Robot’s UR5 with CB2 controller. The system is built using Unity Game Engine and and can be operated from virtual reality.
The project source code can be requested here.