Reinforcement Learning-Based Control of Single-Track Two-Wheeled Robots in Narrow Terrain

The single-track two-wheeled (STTW) robot has the advantages of small size and flexibility, it is suitable for traveling in narrow terrains mountains jungles. In this article, a reinforcement learning control method STTW robots proposed driving fast terrain with limited visibility line-of-sight occlusions. scheme integrates path planning, trajectory tracking, balancing single framework. Based on method, state, action, reward function are defined passing tasks. At same time, we design actor network critic structures use twin delayed deep deterministic policy gradient (TD3) to train these neural networks construct controller. Next, simulation platform formulated test performances method. results show that obtained controller allows effectively pass training terrain, as well four terrains. addition, article conducts comparison prove integrated framework over traditional methods effectiveness function.