Motion Analysis And Control Simulation Of A Bionic Dexterous Hand For Irregularly Shaped Target Grasping

The bionic grasping mechanism consists of a bionic robotic arm and an end dexterous hand,which has a human-like structural appearance and is the most important part of the humanoid robot system for applications such as home service robots and rehabilitation medical robots.In this paper,we propose a new grasping system with UR5 robotic arm structure and design a 16-degree of freedom bionic five-finger dexterous hand capable of grasping irregularly shaped targets,and carry out theoretical and simulation experiments from structural model design,forward and reverse kinematic analysis,operability and singularity analysis,workspace analysis and model optimization program design based on secondary development.The theoretical and simulation experiments are carried out in terms of structural model design,forward and reverse kinematics analysis,operability and singularity analysis,workspace analysis and model optimization program design based on secondary development.First,a set of bionic grasping mechanism was designed by analyzing the limb structure and action characteristics of human in the process of target grasping.A 6-degree-of-freedom UR5 robotic arm is analyzed and used as the arm of the bionic system,and a bionic five-finger dexterous hand with 4 degrees of freedom for the thumb and 3 degrees of freedom for the remaining four fingers is designed.The world coordinate system of the bionic grasping mechanism was analyzed and established,and the forward kinematic models of the robotic arm and the dexterous hand of the bionic system were derived by the flush transformation method and the DH parameter method.Based on this,the inverse kinematic model of the bionic mechanism was derived by using the analytical method,which laid the theoretical foundation for the subsequent study.Second,the Jacobi matrices of the 6-degree-of-freedom robotic arm and the dexterous hand were derived using the differential transformation method and the displacement equation derivation method,respectively.Accordingly,the operability of the bionic system was analyzed.The results show that the optimal maneuverability can be obtained when the sixth joint angle is 0° in the robot arm system under the constraint condition,and the three optimal joint angles are 90°,50°,and 40° in order;by finding the optimal maneuverability of the finger,the thumb structure can meet the joint angles of-10°,-10°,31°,and 45° at the same time with the fingertip joint size of 25 mm.For the other four fingers,the optimal operating degree was achieved when the three joint angles of 85°,100° and 80° and the joint lengths of 60 mm and 40 mm were satisfied simultaneously.The singularity of the bionic system was discriminated,and the mechanism was in singularity when the fifth joint angle of the robot arm was 0° and 90 °;the thumb in the dexterous hand structure had no singularity,and the singular solution existed when the relationship θ1=θ2=0° or θ2=-2θ1+180°,θ1∈[40°,90°]was satisfied for the remaining four fingers when θ3=0°.Then,the workspace of the bionic system was analyzed.Based on the MATLAB robotics toolbox,a visual analysis of the workspace was performed for the 6-degree-of-freedom robotic arm model using the Monte Carlo method.The workspace analysis was performed for each of the five fingers of the dexterous hand by traversing the angles in the kinematic model.Further,a SolidWorks-based secondary development system for the bionic five-finger dexterous hand model was established with the optimal manipulability mapped knuckle length as the optimization index and the workspace as the constraint.With this secondary development system,the ideal structure model can be built quickly for the required workspace with the optimal maneuverability of the dexterous hand.Finally,a joint simulation platform based on ROS and Gazebo was developed for the design of the bionic gripping mechanism.The ROS-based URDF description file was written,and data corrections were made in MashLab software for mass,inertia matrix and other related parameters.The joint model of depth camera,robotic arm and dexterous hand was built using Moveit assistant,and the position joint controller was selected for simulation.Based on the API interface of moveit_commander called by python,an automatic grasping and placing program for different feature objects was written.The grasping methods of the dexterous hand were analyzed,and the simulation experiments of side-grip envelope grasping without changing the posture,side-grip multi-finger tip pinching,five-finger joint side-grip grasping with variable posture and top-grip pinching were carried out to further verify the validity of the simulation model.