Research On Obstacle Detection And Avoidance Strategies Of Construction Handling Robots

As an important equipment in the process of construction industrialization,construction robots are still in the initial stage of development in China.Autonomous obstacle avoidance is the prerequisite for realizing the automatic handling function of construction handling robots.The thesis takes the construction handling robot as the research object,and carries out research on the obstacle detection and obstacle avoidance strategies of the construction handling robot aiming at the defects of incomplete obstacle information obtained by using infrared and other detectors in the traditional method.The paper first summarizes and analyzes the development status of construction robots,and summarizes the domestic and foreign research status of stereo matching technology and obstacle avoidance methods in binocular stereo vision technology.Through literature review and on-site investigation,the research content and method of this article were determined,and the overall strategy of obstacle detection and obstacle avoidance for construction handling robots was completed.Binocular camera calibration and stereo correction are the basis for the entire obstacle detection based on binocular vision.According to the camera imaging model,four coordinate systems in three-dimensional space and image space are established.Through the transformation between the coordinate systems,the relationship matrix between a point in space and pixels on the image can be derived.Secondly,the camera calibration technology is used to complete the extraction of the main parameters of the binocular camera,and on this basis,the stereo correction of the binocular image is completed.Stereo matching is an important link in the entire obstacle detection process,and it is also the most cumbersome step.It is the focus of this paper to reduce the mismatch rate of stereo matching and improve the matching accuracy and efficiency.The main factors affecting the matching accuracy in the local stereo matching algorithm are studied.The multi-feature cost is added in the cost calculation stage.The aggregation matching cost and the occlusion points are eliminated to reduce the mismatch of stereo matching in complex scenes and areas with weak texture performance.Rate,improve algorithm accuracy,and obtain high-quality parallax maps.Using the disparity map obtained by stereo matching,the 3D point cloud reconstruction of the disparity map and the extraction of obstacle depth information are completed.According to the parameters obtained by binocular camera calibration,a 3D point cloud of obstacles is generated.The method of image threshold segmentation is used to extract obstacles in the depth map,and the minimum enclosing rectangle method is used to complete the extraction of obstacle outlines,which realizes the recognition and positioning of obstacles by construction handling robots.An artificial potential field method based on the physics midfield theory is used as the basis of this study to achieve autonomous obstacle avoidance for construction handling robots.Analyze the factors that affect the robot’s motion in the virtual force field of the traditional artificial potential field method,modify the original gravitational field function,and add a heuristic search algorithm—simulated annealing method on this basis to make the handling robot reach the preset target location.The improved stereo matching algorithm and artificial potential field method effectively improve the accuracy of obstacle recognition in the construction scene of the construction handling robot and the efficiency of autonomous obstacle avoidance.It can meet the basic needs in the construction environment.