Research Of Dual Arms Cooperation Technology Based On Machine Vision Technology Based On Machine Vision

Recently,research hotspots have gradually shifted to service robots in the field of robots.The development of artificial intelligence technology has also given robots more intelligence and complex functions.Robots take advantage of visual sensor and manipulators to complete thses action.Single arm operation has shortcomings such as insufficient load capacity,limited range of activities,and so on.Therefore,the research of vision-based dual-arm collaboration is critical.The dual-arm cooperative visual hardware system is built by two KINOVA arms and ASUS Xtion depth camera,and the software is based on ROS.We adopt the DH coordinate method to derivate forward kinematics and sovle the inverse kinematics based on the gradient descent method.In the term of visual positioning,firstly,we complete the monocular camera inside and outside parameters calibration relied on pinhole model and “eye to hand” calibration.Secondly,the target object localization technology based on depth camera and artificial feature April Tags is studied respectively.We design the image segmentation algorithm with strong applicability and robustness.We can Get the pose information of the target object according to segmentation results.In order to verify the correctness of the algorithm in this paper,a simulation test environment is established based on RVIZ platform and URDF model in ROS.In the simulation environment,the master and slave arms collision detection is carried out in the C space of the manipulator by means of the simplification of the manipulator model and the minimum cylindrical envelope.What's more,this paper focuses on the RRT algorithm which is suitable with high dimensional space trajectory planning and do some simulatation in RVIZ.In view of its shortcomings such as blind expansion,weak goal orientation,non-optimal and so on,the strategies that we use include sampling bias target node and updating the parent tree node.In order to solve the trajectory planning of the tight coordination operation,the position and attitude constraint relationship between arms is obtained.Then,RRT algorithm is used to carry out the dual-arm trajectory planning.Finally,in order to verify the correctness and stability of the proposed algorithm,we has carried out experiments,such as grasping book by dual-arm.We analyze the results of the experiments,and validate the reliability and stability of the dual-arm visual hardware and software system.In the end,we point out the current shortcomings and the problems to be studied later.In this paper,the research on the coordinated movement,trajectory planning and visual positioning of the dual-arm system will help the robot to perform the operation accurately,efficiently and quickly.And it also lay a solid theoretical foundation for the application in the actual environment.