Planification de mouvements pour un robot mobile autonome tout-terrain : une approche par utilisation des modèles physiques

Abstract

This thesis deals with the problem of motion planning for an articulated mobile robot moving on an uneven terrain. In this context, the robot is subject to various constraints depending on the dynamics and kinematics of its mechanical system, the shape of the terrain and of the obstacles, and the properties of the contact areas and the physical interactions. The main contribution of this thesis is a two-level planning approach which takes uniformly into account such various features and constraints. The first level is based on a discrete search strategy operating on a subset of the configuration space of the robot, and consists in determining intermediate safe and potential sub-goals to be reached. The second level is based on a continuous motion generation technique allowing to cope with dynamic and interaction constraints. This is achieved locally between each pair of sub-goals by searching for continuous trajectories considering a state space formulation of the motion planning problem and using appropriate physical models.