An extended navigation framework for autonomous mobile robot in dynamic environments using reinforcement learning algorithm

In this paper, we propose an extended navigation framework for autonomous mobile robots in dynamic environments using a reinforcement learning algorithm. The main idea of the proposed algorithm is to provide the mobile robots the relative position and motion of the surrounding objects to the robots, and the safety constraints such as minimum distance from the robots to the obstacles, and a learning model. We then distribute the mobile robots into a dynamic environment. The mobile robots will automatically learn to adapt to the environment by their own experienced through the trial-and-error interaction with the surrounding environment. When the learning phase is completed, the mobile robots equipped with our proposed framework are able to navigate autonomously and safely in the dynamic environment. The simulation results in a simulated environment shows that, our proposed navigation framework is capable of driving the mobile robots to avoid dynamic obstacles and catch up dynamic targets, providing the safety for the surrounding objects and the mobile robots.