Minimum-Time Motion Planner for Mobile Robots on Uneven Terrains

Abstract

This paper presents a method for generating the minimum-time path and motions of a vehicle, taking into account, the relief, the obstacles, the ground surface characteristics, and the vehicle dynamics, including engine and gear.

The terrain is modelled by a triangulation, obtained from the level curves and characteristic points. Monodimensional elements like rivers and roads can be added as polygonal lines, and regions of various surface characteristics are sets of such mono- or bi-dimensional elements, the obstacles being affected by a vehicle mobility equal to zero. The other edges of the model are candidates for generating parts of the optimal path. A A*-like search is used for that. Taking the travel time as the cost function, the velocity of the vehicle is computed from its dynamic model: the power/rpm engine characteristic is approximated by a third degree polynomial from which the driving force is computed. The model of the terrain supplies the required resistant forces due to slope and friction. In climbing motions the best gear is selected automatically by an appropriate algorithm while the downhill parts can be passed either at maximum velocity or kept the same as when climbing. Realistic examples are given which demonstrate the ability of the method for generating the orders required by the pilot: heading angle, gear, and speed.