Mobile Robot Local Path Planning Based On Stereo Vision

With the rapid development of mobile robot technology,local path planning has got great attention.This paper set up a stereoscopic system for autonomous navigation,and the mobile robot can make path planning with binocular camera.The stereoscopic system is divided into hardware system and software system.In the aspect of hardware,a Pioneer 3-AT mobile robot is chosen as the robot platform,and a Bumblebee binocular camera is chosen as the main sensor.APTU-46-17.5 device equipped on the robot makes the camera can adjusts the field of view.A MIO-9290 computer installed on the robot to improve the system real-time computing power.In terms of software system,sensing layer,planning Layer and execution layer were designed based on the Robot Operating System(ROS).The system is rationally structured,highly cohesive,loosely coupled.Binocular camera use two cameras to simulate the 3D feeling of the human eyes,and it can get the 3D information after calibration and stereo matching.A method to road recognition,which is based on Random Sample Consensus(RANSAC)algorithm is designed for traveling along the road.The system can detect obstacles for collision avoidance.The improved RANSAC algorithm is used to lane detection,and the system has feature of real-time and robustness.Finally,the results of the scene recognition are used in local path planning.Based on the behavior-based algorithm,the path planning task of the robot is divided into along the road towards the target behavior and obstacle avoidance behavior.The system realizes a reasonable path in the real-time in a variety of environments with the fused behaviors,behaviors are fused in a subsumption architecture.The experimental results show that the mobile robot system can complete the autonomous navigation through the designed binocular stereo vision in the outdoor environment.