Obstacle Avoidance And Navigation Of Wheeled Mobile Robot

Obstacle avoidance is a very important problem for autonoumous nagivation of mobile robot.The mobile robot uses the distance sensor to obtain the environment information,and the original information is processed and analyzed.This paper is devoted to the research of wheeled mobile robot obstacle avoidance and navigation in reality.The main work of this paper includes:(1)The hardware platform and software platform of mobile robot navigation is built,and the data collection and algorithm verification are carried out on this platform.Mobile robot obstacle avoidance and navigation need a mobile unit to move as a carrier,and the information of the environment is necessary,which is obtained by distance measuring sensor.The laser LMS-200 and wheeled mobile robot Pioneer3-AT could fully meet the above requirements including the algorithm research and experimental verification.(2)The angle potential field method is applied to the obstacle avoidance and navigation of wheeled mobile robots.We use the mobile robot motion direction as the domain to describe the influence of mobile robot obstacle and target point respectively by drag and gravity.With the comprehensive evaluation of the two,we complete the obstacle avoidance and navigation of mobile robot.To a certain extent,this method solves the problem of threshold sensitivity and dead time of similar methods.In addition,in addition to the target position,the angle potential field method does not need any global information,which is not sensitive to the threshold value,so it has good adaptability to the complex changes of environment.(3)We propose a new algorithm ELM-RL which combine reinforcement learning and extreme learning machine.ELM-RL is used in intelligent navigation of mobile robot,this provides a new idea for the obstacle avoidance and navigation of mobile robot.This algorithm not only inherits the advantages of ELM training speed,but also has a great improvement on the training accuracy of ELM algorithm.