Path planning self-learning algorithm for a dynamic chaning environment

The relevance of Human Robot Interaction to complement human skills in a manufacturing environment with industrial robots increases the concerns over safety of human and the robot. It is necessary to identify collision risks and avoid them otherwise production stops may cost a huge amount to the industry. A robot working at manufacturing facility should be able to predict potential collisions and must be able to prevent i.e. react automatically for safe detour around obstacle/human. Currently, industrial robots are able to detect collisions after a real contact but the existing proposals for avoiding collisions are either computationally expensive or not very well adapted to human safety. The objective of this paper is to provide intelligence to the industrial robot to predict collision risks and react automatically without stopping the production in a static environment. The proposed approach using Time of Flight (TOF) camera, provides decision regarding trajectory correction and improvement by shifting robot to a secure position. The application presented in this paper is for safe KUKA robot trajectory generation in peg-in-hole assembly process in the laboratory context