Collision prediction using plan learning in mixed human–robot work cells

In mixed human-robot work cells the emphasis is traditionally on collision avoidance to circumvent injuries and production down times. In this paper we discuss how long in advance a collision can be predicted given the behavior of a robotic arm and the current occupancy of both the robot and the human. The behavior of the robot is a sequence of predefined operations that constitute its plan, each one with a given trajectory. However, we do not know the exact trajectory or the plan a priori. Under the assumption that the plan has a cyclic character, we propose an approach to learn it in real time from state samples and use the resulting model to estimate the time before a collision. The pose of the human is obtained by a multi-camera inference application based on neural networks at the edge to preserve privacy and enforce scalability. The occupancy of the manipulator and of the human are modeled through the composition of segments which overcomes the traditional ``virtual cage'' and can be adapted to different human beings and robots. The system has been implemented in a real factory scenario to demonstrate its readiness regarding both industrial constraints and computational complexity.