mobile robot path planning using a new ga-based method with variable chromosome length

being one of the major research fields in the robotics discipline, the robot motion planning problem deals with finding an obstacle-free start-to-goal path for a robot navigating among workspace obstacles. such a problem is also encountered in path planning of intelligent vehicles and automatic guided vehicles (agvs). traditional (exact) algorithms have failed to solve the problem effectively since it is proven that the complexity of the problem is np-hard. consequently, several heuristic algorithms have been developed for solving the problem, among which the genetic algorithms (ga) evolutionary approach has been increasingly utilized. in this paper, a new ga-based method has been developed to improve the drawbacks of current ga methods regarding fixed chromosome string size and limitations in the crossover and mutation operators. in this method, first the workspace is tessellated via the delaunay triangulation, and a set of paths are created using the powerful rapidly-exploring random trees (rrt) and smoothed. as a result, a population of variable-length chromosomes is obtained, which is a novelty in the field. next, these solutions are improved further by using new selection, crossover and mutation operators, in which the triangulation plays a pivotal role. extensive simulations of the algorithm showed that the obtained path lengths are averagely 25% shorter than those found by the pure rrt method.