Research On Key Technology Of Dexterous Grasping With Teleoperation Of 6-Axis Lightweight Robot Arm

In unstructured environments,it is a mainstream direction to complete remote operations with the use of robot teleoperation based on human-robot interaction technology.Teleoperation enables robots to obtain strong environmental adaptability and learning ability,in order to accomplish more complex tasks under human remote processing.The robot arm teleoperation and dexterous grasping include three phases of operation: fast approaching towards the target object,automatically grasping posture adjustment and dexterous grasping.The kinematic similarity mapping between man and robot arm in teleoperation and dynamic force distribution of dexterous grasping are the two key techniques for this research.This paper conducts both theoretical and experimental research on the above two key technical issues.The main work in this paper is listed as follows:1)A 6-axis lightweight robot arm teleoperation mapping and optimization method is proposed.In the approach of the robot arm teleoperation towards the target object,since the human upper limb and the robot arm have different joint configurations,we respectively define the principles of human-robot motion similarity in both Cartesian space and joint space,and make full use of the characteristics of robot arm redundancy.An algorithm based on redundant degree of freedom decomposition for manipulators is proposed.A teleoperation mapping method that can realize Cartesian space and joint space coupling control is proposed and applied to the 6-axis lightweight robot arm teleoperation system.Compared with the existing methods,the teleoperation attitude mapping algorithm proposed in this paper is simpler in operation,and better maintains the control intuitiveness and operational similarity of the follow-up motion guided by the human body posture.2)A 6-axis lightweight robot arm dexterous grasping dynamic force distribution method is proposed.In the dexterous grasping stage,the second-order cone programming problem is solved as a semi-definite programming problem for the current mainstream dexterous grasping dynamic force distribution method.The computational complexity is high.This paper draws on the idea of penalty function method.A proper and efficient method for constructing the dynamic force distribution problem is proposed.The algorithm proposed in this paper is superior to the mainstream semi-definite programming method in terms of execution time and number of iterations.3)A 6-axis lightweight robot arm teleoperation and dexterous grasping experimental platform is constructed.Using Kinect as a human body attitude measuring device,a teleoperation motion control software is developed.A target corner search algorithm is proposed to realize binocular camera corner reconstruction,and complete target object pose measurement.The binocular camera is calibrated,and a hand-to-eye calibration between the camera and robot arm is implemented,in order to calculate the conversion relationship between the binocular camera and the robot arm coordinate system.4)The teleoperation real-time follow-up experiment based on the 6-axis lightweight robot arm teleoperation mapping and optimization method,and the grasping-quality experimental research based on the 6-axis lightweight robot arm dexterous grasping dynamic force distribution optimization method are implemented.In order to verify the real-time following performance of the robot arm based on the attitude mapping method proposed in this paper,the real-time following accuracy of the robot arm teleoperation is measured.In addition,in order to evaluate the grasping quality of the dexterous grasping dynamic force distribution method proposed in this paper,the impact of the penalty factor on the quality of the dynamic force distribution method is analyzed from the quantitative perspective,and the target corner point search accuracy and the positioning accuracy of the grasping posture adjustment stage are evaluated are evaluated.From the perspective of simulation experiments,it is verified that the proposed algorithm has better computational performance than the mainstream semi-definite programming algorithm,with the feasibility of the robot arm teleoperation dexterous grasping application.