Dynamic Obstacle Avoidance of Mobile Robots Using Real-Time Q-learning

As the field of autonomous navigation has been actively researched, the importance of route search is increasing. In particular, the field of autonomous navigation using reinforcement learning is being intensively studied. However, this requires very complex algorithms and high cost. Previous studies have shown that path planning can also be performed with Q-learning, a lightweight reinforcement learning algorithm, with proper selection of the exploration strategy. In this paper, we show that real-time Q-learning can be used for path planning and dynamic obstacle avoidance of mobile robots in various environments. And in a follow-up study, we plan to apply real-time Q-learning to real mobile robots rather than simulations.