Design Of The Robot Control System Based On Monocular Vision

With the development of the economic and social,the standard of people's living has been improved,more and more robots have entered people's life.However,the existing robots usually have imperfect design of the sensor system and irrational path planning algorithm,resulting in the robot in some unknown work environment,unnecessary collision,or even an accident.Consequently,a robot control system based on monocular vision is designed in this thesis.The monocular camera equipped with the robot to collect the external environment information is used in the system;the two-dimensional image information is converted into three-dimensional information in the world coordinate system,so that the environment information of the robot work is modeled,and guided the robot trajectory planning.The specific research contents and the results obtained are as follows:Firstly,the image of the collected should be preprocessed.Image preprocessing includes image grayscale and image denoising,so that the collected images can meet the requirements of precise positioning.The key problem of target location is the extrinsic matrix calculation of the camera.The matrix calculation of the camera is calculated from the camera's intrinsic matrix combined with the measured camera parameters.The Harris algorithm is used to detect the matching feature points.The three-dimensional coordinates in the world coordinate system can be calculated according to the correspondence between the established image coordinate system and the world coordinate system.Secondly,the Free Space Approach is used to construct the world model according to the three-dimensional environmental information,as to obtain a complete environment map model.The A*(A Star)algorithm is used to search the map for paths.The A*algorithm is an efficient global path search algorithm,which can obtain the shortest path for the entire map to complete the path search in a short time.However,there are some limitations in the A*algorithm,PRM(Probabilistic Roadmap Algorithm)algorithm is combined to improve the A*algorithm,and get a more efficient path search method.Finally,by combining the target positioning module with the trajectory planning module,a comprehensive experiment is carried out to verify the feasibility of the whole system.In summary,a robot control system based on monocular vision researched in this thesis can obtain the environmental information in unknown environment and realize the robot trajectory planning.