Study Of Gesture Recognition And Grasping Technology Of Robot Demonstration System

Multi-finger dexterous hand can do more subtle and complicated task than common gripper and is playing an increasingly important role in domestic services and industrial production. More and more researchers around multi-finger dexterous hand have been carried out. This dissertation presents a thorough research on the Gesture recognition, motion mapping, Tactile feedback and grasping quality measurement, and also puts forward some views and methods about the hand.The first part of this dissertation focused on the shortage of the single LeapMotion gesture recognition and the necessity of introducing the second LM for multi-sensor fusion. Then we determined the installation position of the sensor, analyzed the working space and coordinate transformation and design a corresponding bracket and clip for installation. We analyzed the LM observation model and hand motion model to establish a multi-sensor fusion model. Finally, we set up a complete double LM fusion system, and completed the visualization output and experimental verification.In order to study human experience, hand movement should be mapped to the dexterous hands. We completed the SDH positive kinematics solution and used SVM to classify hand gestures and pose mapping. By solving the inverse kinematic solution of SDH and end effector, we completed the point to point mapping.We use the tactile feedback to increase the grasping reliability, decrease the grasping force and provide the contact force information. We solved the force closure problem of grasp configuration and grasping quality measurement.Finally, we create robot demonstration simulation platform based on the Robot Operating System(ROS) at last. In the part of virtual experiments, we complete two virtual experiments: pose mapping and point to point mapping. The availability of robot demonstration system is verified.