Reactive path planning with collision avoidance in dynamic environments

We present an approach to path planning for remote surveillance robots in dynamic environments. The method has to increase robot autonomy. The problem is considered at two levels: using information from an a priori environment map, a path which avoids all the static obstacles is planned. Next, when following the path, the robot updates the map and adjusts the path with sensor information avoiding unexpected obstacles. The updating is done at quasi regular periods of time. We verify the efficiency of the method through simulation results.