Off-road Terrain Path Planning for Differential Steering Vehicles Based on Artificial Potential Field Gradient*

Motion planning of unmanned tracked vehicle (UTV) involves not only highly nonlinear vehicle dynamic constraints, but also challenges from complex rough off-road terrain with dynamic and static obstacles. In view of the UTV motion planning, this paper systematically consider vehicle kinematic model, off-road terrain and obstacle factors, by introducing different types of potential energy function models corresponding to two kinds of obstacles and ground. Regarding potential field gradient as a condition of optimization, a multi-stage optimization Channel Path Planning (CPP) is designed to avoid the impracticable path formulated by Linear Path Planning (LPP), and converted to nonlinear programming problem through the pseudo-spectral method. By means of multi-stage optimization strategy, the computation of the original optimization algorithm can be reduced while ensuring global optimal approximate solution. Simulation experiments show that this method can catch feasible paths effectively in the process of path planning even with respect to vehicle dynamic constraints, obstacle and off-road terrain.