Path Planning Of Mobile Robot Based On Neural Network And Genetic Algorithm

Path planning of Mobile Robot is the core of Robotics research, also it is the key technology to bring Robot's automation and intelligence into effect. In order to work independently in the environment, the Robot is considered to have the ability of perceiving workspace and controlling the movement of itself. The perception of Mobile Robot is sensing the information of work environment by the use of its own carrying sensors, while the motion control of Mobile Robot is utilizing the acquired data information and the control rules when it encountered specific obstacles to determine their own velocity and orientation, that is, the path planning.First, the application of Neural Network and Genetic Algorithm in the Robot's path planning was analyzed respectively in this thesis. Generally, Neural Network were used to fuse the Mobile Robot's sensing data for creating the map of their working scope, while Genetic Algorithm had a good performance in path planning when the workspace was globally known, but in the face of unknown environment it was often powerless. To this end, a technique was proposed in this thesis for realizing the path planning in unknown environment by the combination of Neural Network and Genetic Algorithm. In this method, the Mobile Robot's working environment was assumed to be two-dimensional, obstacles with arbitrary shape, number and position, a set of points in the workspace constituted the planned path. Mobile Robot's motion control rules was decided by the Neural Network, first, the Robot perceived its surrounded working environment with the five front-end configured sensors, then, the perceived obstacle information and the distance/orientation information of the target was taken as the network's inputs, the weight training of the Neural Network was completed with the Genetic Algorithm, finally, we got the outputs of the trained network as the driving force to achieve the path planning in the unknown environment.It can be seen from the simulation results that by using this path planning method, our Mobile Robot can not only avoid obstacles effectively, but also move toward the target and eventually reach the target, thus the path planning in unknown environment is accomplished.