Towards an optimal avoidance strategy for collaborative robots

Abstract

Abstract Collaborative robots represent a game changer in manufacturing for their ease of use, the reduced need of safeguarding hardware and, consequently, their extremely fast payback time. However, most of the collaborative robots available on the market are power and force limiting (PFL) devices. The main disadvantage of this type of collaborative operation is that the robot is forced to stop when a collision occurs, as the only way the robot is aware of the presence of the human is through its embedded torque or motor current monitoring algorithms. Albeit tolerable from a safety point of view, these collisions might dramatically reduce the performance of the robot in terms of productivity, ultimately jeopardising the economic attractiveness of a collaborative workstation. This paper introduces an avoidance strategy that suggests the robot alternative paths to be traversed, that are both collision free and optimal in terms of minimum traversal time. The control strategy makes use of a depth camera in order to enhance robot perception of the environment. Moreover, by properly exploiting information coming from these sensors, the control strategy itself is able to communicate to the robot the best decision to take with respect to the presence of one or more human operators. The method is experimentally validated on a Universal Robots UR5.