Dynamic Trajectory Planning Of Collaborative Robots Based On Human Body Obstacle Detection

Robot is the representative of intelligent equipment and robot technology is the strategic support technology of a country.Robot is important to modern manufacturing,national defense security and social development,so it is regarded as one of the core technologies of the new technological revolution.At present,collaborative robot which can collaborate with people and complete complex dynamic tasks caught the attention of researchers.Traditional industrial robots generally use fences to isolate robots from workers,while collaborative robots can be in the same working area with workers[1].Ensuring the safety of the human-robot system is essential for robot when sharing the same physical space with people.For the collaborative robot,it is a basic safety protection function to realize the collision warning during the movement of the robot and react quickly.In the environments of human-robot coexistence,the activities of workers may hinder the normal operation of robots.Robots will suspend the working process when the collision occurs with the human body or other obstacles,or even worse,if the obstacle is always present,robots will repeatedly reduce speed or even stop working to ensure safety during the operation,and the working efficiency of robots will be greatly affected.Based on the key project of the Chinese Academy of Sciences,"Key Technology and System Development of Next-Generation Industrial Robots"(Project No.KGZD-EW-608-1),this thesis focuses on the security issues in the human-robot collaboration scenario.Using Kinect as the environment-aware sensor and using the UR5 robot as the experimental object,this thesis develops a dynamic trajectory planning method for collaborative robot.The main research work includes:(1)Compared with the traditional industrial robots,this paper analyzes the research development history,technical advantages and application prospects of the collaborative robots,and focuses on the domestic and international research status of the collaborative environment perception and online trajectory planning.The research significance of the dynamic trajectory planning of the manipulator is discussed and the scientific problems are refined.(2)Using the UR5 robots as the research object and verification platform,the kinematics analysis of the collaborative robot is carried out.The forward and inverse kinematics models are established based on the DH parameter method.The model is simulated by the MATLAB robot toolbox.Jacobian matrix of the UR5 robot is obtained by the comprehensive method,and the relationship between the speed of the end effector of UR5 robot and the speed of each joint is established.(3)The working principle and technical application of Kinect camera are analyzed and the mapping model of Kinect depth image to RGB image is established.The internal parameter calibration of Kinect camera is completed by Zhang Zhengyou calibration method.This paper proposes a new kinect camera external reference calibration method based on Hough transform and least squares method,and the external parameter calibration of the Kinect camera is completed by extracting and analyzing multiple sets of point data of the end effector in the camera coordinate system and the base coordinate system.(4)Based on the Kinect library file,the second development of the human detection program was completed,and the position detection of the 26 joint points of the human body in the Kinect camera coordinate system and the image display of the human skeleton model based on OpenCv were realized.Based on the message of the UR5 robot,the position and pose of the end effector in the base coordinate system is obtained.On the basis of this,the human-robot distance online detection is realized by homogeneous coordinate transformation.(5)According to the complexity and unpredictability of human obstacles in human-robot cooperation scene,an APF-RV dynamic trajectory planning algorithm is proposed,which generates online trajectory based on the position and motion speed of the human body and the end effector of the robot in the workspace,and the dynamic collision avoidance planning for human obstacles in different motion states is realized.Further physical experiment is conducted by UR5 robot to verify the feasibility and effectivity of the designed real-time collision avoidance algorithm.Finally,the research work of this thesis is summarized,and the direction to be further studied is put forward.