Continuous - Curvature Path Planning forMultiple

In this article, we consider path planning for car-like robots in a new way, adding a continuous-curvature condition to the paths. As the optimal paths are too complex to be used, we deene a type of continuous-curvature paths inspired of the Dubins' curves. These paths contain at most seven pieces, each piece being either a line segment, a circular arc or a clothoid arc (a clothoid is a curve whose curvature is a linear function of its arc length). The curvature along these paths is a piecewise linear function of the arc length. Then, a local planner (i.e. a non-complete generator of non-colliding paths) that makes use of these paths will be described. At last, we present how to combine this local planner and the Probabilistic Path Planning (PPP) method to obtain a single vehicle global planner. This planner operates in two phases: the learning phase, during which the free connguration space is approximated using a graph, and the query phase, which is mainly a graph search. We also present how to obtain a multi-vehicles path planner. Praxit ele programme on urban public transports. Abstract In this article, we consider path planning for car-like robots in a new way, adding a continuous-curvature condition to the paths. As the optimal paths are too complex to be used, we deene a type of continuous-curvature paths inspired of the Dubins' curves. These paths contain at most seven pieces, each piece being either a line segment, a circular arc or a clothoid arc (a clothoid is a curve whose curvature is a linear function of its arc length). The curvature along these paths is a piecewise linear function of the arc length. Then, a local planner (i.e. a non-complete generator of non-colliding paths) that makes use of these paths will be described. At last, we present how to combine this local planner and the Probabilistic Path Planning (PPP) method to obtain a single vehicle global planner. This planner operates in two phases: the learning phase, during which the free connguration space is approximated using a graph, and the query phase, which is mainly a graph search. We also present how to obtain a multi-vehicles path planner.