The Research Of Robot Path Planning Based On Ga

Path planning is a kernel problem of robot technology area, and is also an important aspect of robot study in artificial intelligence. The main task of path planning is figuring out a collide-free path at least cost from the known start position to known goal position in the robot environment. In environment only with static obstacles, it is easy to find an optimized path, while in environment with both static obstacles and dynamic obstacles, it is more difficult to find a feasible optimized path. Main contents in this paper include followings.Two layered genetic algorithm mechanism is brought up as global parallel optimize searching tool to find optimized path. The first layer genetic algorithm is responsible for static obstacles avoidance, including the existing obstacles in the environment and the new static obstacles appearing in the process of path planning. The second layer genetic algorithm answers for dynamic obstacles avoidance, which is based on the first layer optimized path mechanism, and finally marks out the optimized path for avoidance of all static and dynamic obstacles in theenvironment. Different fitness functions are adopted in the two layers, optimize objects of fitness function in the first layer are path length, corner smoothness and clearance, while the optimize object of fitness function in the second layer includes whether in the safe area or not, whether dynamic avoidance or not, and the path length.Accompanied with optimized path planning, predigest operator is applied to wipe off redundancy nodes, smoothness operator is used to increase smoothness of path angles, and amend operator is adopted to assist creating feasible path in complicated environment. Optimized operations speed up creating optimized path.According to the ratio of feasible paths and infeasible paths, crossover and mutation are adaptive adjusted in order to improve constringency of algorithm mechanism.Using MATLAB to build up simulate platform, optimized paths in evolutions are demonstrated in static state and the moving process of obstacles avoidance is also displayed in environment in dynamic state, including static obstacles and dynamic obstacles.Finally, different optimized paths under different parameters are compared; fitness values of various population sizes are statistically analyzed and best optimized paths and worst optimized paths in different environments are compared as well.