Safe motion planning for car like vehiclesTh

This paper deals with the problem of planning and controlling the motion of non holonomic vehicles in a dynamic world. We present a hierarchical system with two main levels: at the higher level, a motion planner determines a coarse motion plan for each controlled vehicle according to the non collision and time optimality criteria; at the lower level, a motion controller distributed between the diierent vehicles reenes and executes the previously computed motion plan. The contribution of this paper is twofold: (1) we present the motion planner whose purpose is to determine a nominal motion plan for each controlled vehicle. This plan is made up of a nominal geometric trajectory and of an associated set of time constraints of the type \reach location l at time t". The motion planner makes use of the very structure of the intersection both to estimate the behaviour of the other vehicles and to plan the motions of the controlled vehicles. We have developed a method combining a prioritization approach with a path/velocity decomposition. This means that the motion planner makes use of two complementary planners: the geometric planner whose purpose is to generate a smooth maneuver free nominal trajectory for each vehicle and the temporal planner which determines a sequence of time constraints to be associated with each trajectory. In order to deal with the uncertainty introduced by the other vehicles, the motion planner operates periodically on a given interval of time. (2) we describe the motion controller which is concerned with the on line execution of the motion plan. The motion controller operates on a discretized model of the local environment of the vehicle. This model is used to adapt the behaviour of the vehicle to the characteristics of its actual environment (to avoid the other vehicles in particular). This is achieved using a potential elds approach allowing the vehicle to follow as closely as possible the given nominal trajectory and to satisfy as much as possible the associated time constraints.