Path planning approach based on probabilistic roadmap for sensor based car-like robot in unknown environments

This paper describes a new motion planning approaches based on probabilistic roadmap (PRM) algorithm for path planning of car-like mobile robot in unknown environments. Although there are already some approaches which stem from PRM and deal with the path planning problems in this field, the main flaw is the high costs of the robots´ paths, especially in the case that the environments are complex. The planning method in this paper makes a development of those current approaches. The improvements are mainly at two points: (1) the local free nodes regions are dwindled in order to decrease the cost of the path; and (2) the new algorithm creates another group of configurations around the robot within the sensors´ detection distance in each step if the number of current nodes is smaller than an initially settled value, so that the narrow passages are not be ignored by the robot.