Abstract

As human and robot collaboration become more intimate, and more intelligent robots are being used in both healthcare and manufacturing environments, the safety of humans working closely with companion or co-robots is becoming a challenging issue. As a result, there is a great need for developing improved control algorithms and better guidelines to allow humans and robots to safely work together in a synergistic fashion. The nature of human variability requires the robot to be able to learn and adapt to these changes while maintaining safety in the presence of an often unpredictable human worker. This paper proposes a novel solution to create safe and fault-tolerant motion planning and control of industrial robots for safe human-robot cooperation. First, human motion in a collaborative task with a robot was recorded with a ten-camera motion capture system. Then, a novel algorithm developed on for fault-tolerant motion planning and control of industrial robots to increase the human safety in human and robot collaboration. It has been done by characterizing human motion patterns in real-time while preventing collision between human and robot and minimizing harmful sudden motions when a failure occurs in robot actuators. The proposed algorithm benefits from using convex optimization, disjunctive programming and receding horizon concept to work in real-time. Simulation studies has been developed in ROS and Gazebo softwares and results reveals the performance of the algorithm to solve the simultaneous fault-tolerant and collision-free motion planning problem in industrial robots. Using the proposed algorithm will minimize velocity jumps of robot due to sudden actuator failures and avoid collision with human worker while maintaining acceptable levels of productivity.