Deep Learning and Game Theory for AI-Enabled Human-Robot Collaboration System Design in Industry 4.0

In this paper, an optimization problem in a human-robot collaboration system is addressed. In Industry 4.0, mobile robots and stationary human workers work simultaneously in the same working environment. The robots have to determine the optimal path to move while avoiding colliding with the human workers and other robots. The challenge is for the robots to determine their optimal paths while avoiding collisions with both human workers and other robots. Traditional Deep Reinforcement Learning exhibits poor performance in this complex scenario, primarily due to slow convergence. To mitigate this issue, we propose a hybrid approach that combines Deep Reinforcement Learning with game theory. In the algorithm, first, a heuristic approach is used to evaluate the potential for collisions between robots. If no collision risk is detected, the Deep Reinforcement Learning algorithm is used to determine the optimal path for each robot in serial with the working environment updated iteratively. If the collision risk exists, a cooperative game is formulated within the Deep Reinforcement Learning algorithm framework to resolve the collision concern. The numerical results prove the superiority of the proposed algorithm in solving the distributed path planning problem for each robot compared with the state-of-the-art.