Research On Autonomous Obstacle-Surmounting Of Swing-Arms Robot Based On Vision

In all parts of the world,natural disasters,terrorist activities and various emergencies happen frequently.We need to use mobile robots to replace human beings in these difficult,dangerous and harmful environments.When mobile robots are engaged in reconnaissance and emergency response tasks,autonomous obstacle surmounting ability is one of the basic intelligent behaviors that robots need to accomplish such tasks as dangerous environment exploration and so on,as one of the most widely used environmental modeling and obstacle detection methods,vision is being applied to all kinds of indoor and outdoor mobile robots.Aiming at the most common and complex staircase obstacle model in indoor environment,we have carried out vision based environmental feature recognition,geometric parameter acquisition and swinging robot autonomous obstacle surmounting research.This paper analyzes the principle of binocular vision image and depth information,according to the environmental feature recognition task,study the processing method of 3D point cloud,the processing algorithm for 3D point cloud data structured barrier design,presents a projection space division method based on the idea of grid.On this basis,the normal vector computation of point cloud data is carried out,and then the plane area segmentation is performed according to the normal vector,and the straight line feature is extracted,and the obstacle parameters of stairs are obtained,and the vision based environmental feature recognition is realized.The kinematic modeling of the swing arm robot is analyzed.The obstacle obstacle abilities of the robot are analyzed,including the maximum height of climbing a single step and the ability to climb a staircase.Based on this,aiming at different steps and stairs,Autonomous obstacle robot carried out action planning.The model of the robot's center of mass is established.The static stability of the robot is analyzed by the method of stable cone.The simulation is analyzed by MATLAB.The hardware and software construction of the independent obstacle avoidance experiment platform was completed.Experiments were carried out to ascend the steps and stairs automatically,which verified the rationality and feasibility of robot's passing through,obstacle surmounting stability and obstacle crossing motion planning.