Experimental Evaluation of Collision Avoidance Techniques for Collaborative Robots

This paper presents the implementation of an obstacle avoidance algorithm on the UR5e collaborative robot. The algorithm, previously developed and verified in simulation, allows one to modify in real time the trajectory of the manipulator with three different modalities to avoid obstacles. Some test cases with fixed or dynamic obstacles affecting the robot's motion were first simulated and then experimented on. The paper describes the hardware/software architecture of the robotic system: an external controller is realized by a standard PC that communicates with the robot controller by a TCP/IP protocol; algorithms and data processing are executed by Python/Matlab software that guarantees a duty cycle of at least 100 Hz. The error analysis between simulated and real data allows one to conclude that the developed algorithms revealed to be effectively applied to a real robotic system, showing behavior similar to what is expected by simulations.