A Novel Gradient-Based Motion Planning for Single-Track Two-Wheeled Robots in Complex Environments

Single-track two-wheeled robots with the excellent characteristics of lightweight, trafficability and maneuverability exhibit great potential in goods transportation, field patrol and exploration in narrow environments. Motion planning as an indispensable component acts an essential role in autonomous navigation systems of single-track two-wheeled robots. This paper presents an efficient and robust motion planning framework for this type of robot in unknown and unstructured environments. Firstly, an initial trajectory is generated without regard to obstacles. Then, we obtain the lattice graph to search a feasible path based on the simplified kinematic bicycle model. The collision penalty function is immediately established to push the initial trajectory far away from obstacles as long as it collides with obstacles. The time allocation can be lengthened if the trajectory cannot ensure dynamical feasibility. Simulation experiments demonstrate the proposed motion planning has robustness and high-performance.