Non-specific Structure Manipulator And Robot Pose Mapping For Precision Operation

The robot multi-finger dexterous hand as the robot end operator has the characteristics of multi-degree of freedom and multi-finger coordination which is one of the effective ways to improve the operation performance and operation level of the traditional robot.Using teleoperation taking people as part of the system to participate in the dexterous hand grab control can improve the operational performance and operation level of the dexterous hand and use the human grasping operation experience to carry out high-level task planning and command design and the hand.completes the task execution by smart.The data-glove is an interface device that can realize human-computer interaction well.But the mapping from the human hand to the robot is particularly important because of the different kinematic structure of the human hand and the robot.This paper summarizes the research situation of manipulator-to-manipulator operation mapping at home and abroad.Based on data gloves and three-finger manipulators and the following contents are studied with reference to the existing finger mapping methods and human gesture recognition for manipulating objects.The kinematics analysis of the virtual human hand,the positive and negative kinematics model verification results show that the four-finger and thumb kinematics model represented by the index finger is established correctly.The main research contents include:using the D-H matrix transformation method,the forward kinematics modeling of the four fingers represented by the index finger,the forward kinematics modeling of the thumb,the inverse kinematics modeling of the index finger and the thumb,the index finger and the thumb.Verification of the forward and inverse kinematics model.A kinematic analysis of the Barrett hand manipulator was performed.The motion trajectory space of the virtual human hand is simulated.The simulation results show that the kinematics modeling of the virtual human hand is correct,and the virtual trajectory space is very similar to the real human hand motion trajectory space.The main research contents include the simulation of the trajectory space of the index finger in the natural state,the left index of 15°,and the right exhibition of 15° using MATLAB software.The three fingers of the middle finger,ring finger and little finger are in the natural state.The simulation of the motion trajectory space simulates the workspace in the side grip and the positive grip posture of the thumb,and simulates the positive grip and the side grip of the five-finger collective.In addition,the thumb and index finger under the positive grip and the side grip state.The spatial relationship of the motion trajectory was also simulatedThe manipulator teleoperation platform was built and the motion mapping method from manual to robot was studied.The main research contents include:data processing of data gloves by serial communication,and manipulator teleoperation control program design based on ROS platform.The joint-to-joint mapping from the human hand to the Barrett hand three-finger manipulator was designed and implemented.In the virtual ellipsoid mapping method,the optimal contact point between the robot and the virtual ellipsoid is selected by the grab quality evaluation based on the grab matrix.A mapping method for two-finger gripping objects is designed.At the end of the design,the MLP multi-layer perceptron-based gesture classification algorithm is designed to realize the recognition of the two gestures of grabbing the ellipsoid object and the two-finger gripping object,and realizing the fusion of point-to-point mapping and virtual object mapping.Extending the scope of the mapping method.