On Collision Detection Method Of Human-Machine Cooperative Robot

With the development of industrial society,people's demand for personalized and customized products has gradually increased.Advanced manufacturing has gradually become the future development trend of manufacturing,and the serious deficiencies of the existing traditional industrial robots' mobility flexibility,unstructured environment adaptability and intelligence has been gradually exposed.In order to solve these problems,human-machine cooperation robot emerges as the times require,in the application of human-machine cooperation,robot and human need to share the working space.In order to protect the safety of human in collaborative work,it is of great significance for the research of robot collision detection.Based on the six degree of freedom cooperative robot,a new collision detection method is proposed in this thesis,the main work and innovations are as follows:Kinematics modeling of cooperative robot.The robot kinematics model is the basis of controlling robot motion.According to the characteristics of cooperative robot mechanical structure,this thesis uses the D-H method and the analytical method to model the forward and inverse kinematics models of the robot,and uses simulation to verify the effectiveness.Dynamic modeling of cooperative robot.The dynamic model of robot can establish the relationship among its pose,motion state and joint moment.In this thesis,Newton Euler method is used to establish the dynamic model of the robot,excitation in discrete Fourier form is selected to design the joint excitation trajectory,and the minimum condition number of the dynamic parameter matrix is taken as the optimization objective to solve the excitation parameters.The minimum parameter set is estimated by collecting the data under the excitation.Research on collision detection method of robot.In order to distinguish whether the external force is caused by human active contact or collision,this thesis proposes a detection scheme based on frequency domain analysis.A sliding observation window is set for the moment value of the external force,and the high frequency part corresponding to the moment mutation is selected for derivation and filtering.The magnitude of the result indicates the intensity of the collision.Experiments show that the new algorithm can effectively identify the high-frequency torque value when collision occurs,and can effectively distinguish the slow change of external torque caused by active touch and load change.The combination of frequency domain analysis and traditional fixed threshold methods can make a judgment between the normal operation of the robot,human intervention or collision,and guide the response strategies of the robot to different situations.