Motion Control Of Arm-gripper System And Research On Stable Grasping

With the development of technology and improvement of people’s life quality,robots come to people’s life more and more quickly. As the execution devicesinteracting with the environment, the robot arm and gripper is an important way ofrobot to operate objects, what’s more, it’s a kind of prerequisite for robot to havemore complex applications.A reliable, robustness, adaptive grasping system is designed and implementedin this paper by combining KUKA arm and a self-designed two-finger gripper withvisual, tactile sensors, infrared distance sensor, proximity tactile sensor and othersensors. Binocular vision supplemented monocular camera and infrared distancesensor are used to match goal and position precisely. Hybrid force/position controlmethod is applied to control the grasping system and stable grasping is realized bymulti-sensor fusion technology.The main content of this thesis is as follows：(1) Build the robot crawl operating platform, KUKA manipulator and a two-finger gripper are the mechanical body. The system is divided into visual module,manipulator motion control module and gripper grasping module. In the softwarepart, the system runs in ROS on Linux platform and uses Topic session mechanism toachieve the integration and communication between the different modules.(2) Binocular vision, monocular vision and infrared range sensor comprise therobot vision system. The main positioning method is binocular vision position, if itfails, monocular visual and infrared distance sensor are used to auxiliary. By thisway, the robustness of the system is improved largely. SIFT features algorithm isapplied to match and stereo vision technology is used to locate target, what’s more,an object recognition and localization experiment is done and the results prove themethod’s validity.(3) Analysis the kinematics of the robot arm, realize force/position hybridcontrol by fuzzy adaptive PD control method. The process is simple, robustness andcan meet the system needs.(4) Multi-sensor data fusion technology is employed to implement stablegrasping. The improved PSO algorithm is applied to realize adaptive weightedfusion in homogeneity sensors. Rough set and SVM are used to fuse heterogeneous sensors. What’s more, we conducted an experiment to valid above methods, and theresults show that the method can greatly improve the grasping success rate andenhance system performance effectively.At last, the content of this thesis is summarized and further feasible researchproblems are discussed.