Research On The Key Technology Of Robotic Compliant Assembly Control System For Explosive Components

Components assembly is the key process to determine product quality in manufacturing industry.Robot technology is more and more widely applied in industrial production to achieve "machine subsitution".However,robots have not yet been widely used in the field of assembly because of the constraint complexity of assembly tasks and the poor flexibility of robots.Especially for the assembly of explosive components,the control of contact force is very complex because the components are explosive and fragile.The process requirements for safety and reliability are more stringent.At present,there is no domestic application of robots and related technologies for assembly of explosive components,because the assembly technology has not been focused on safety constraints.These assembly tasks are mainly completed manually.The quality of assembly depends entirely on human experience,physical and psychological state.In this way,it is not only inefficient,but also may lead to dangerous accidents that may cause casualties.Therefore,considering the strategic requirement of independent development of national defense industry for high-end equipment,this paper studies the key technologies of robotic compliant assembly control system for explosive components.This paper is aimed to make the robot complete the assembly tasks of explosive components safely and effectively while guaranteeing safety constraints,and solve the core technical problems of robots in the assembly control,providing theoretical support and technical path for the safe,efficient and intelligent assembly of explosive components.Firstly,the assembly process of explosive components is analyzed in detail.The geometric and mechanical contact constraints are sorted out,and the mechanical causes of assembly wedging and jamming are analyzed and summarized.In order to meet the requirements of safety and robustness in the assembly process,a complaint assembly control system for explosive components under safety constraints is constructed,and an assembly strategy containing four basic motions is designed.The complaint assembly control system mainly consists of four parts:target detection and positioning module,motion teaching module,learning variable impedance control module and assembly strategy module.In order to control the assembly force more flexibly and solve the problem of frequent switching force control during assembly process,a compliant assembly control algorithm based on end-force is proposed.The desired motion and force of the manipulator are both completely generated and controlled by the force at the end-effector of the manipulator.Secondly,aiming at the positioning problem of energetic components,the recognition and location algorithm of explosive components based on Kinect depth camera is studied.Based on the point cloud data of the depth camera,the detection and pose estimation of the explosive components are realized using the point cloud segmentation method.For the aim of teaching the robot assembly motion skills and ensuring the robustness while satisfying the constraints,a probabilistic motion teaching method under safety constraints is proposed.The probabilistic encoding and reconstructed generalization of teaching data are carried out by using Dynamic Time Warping algorithm(DTW),Gaussian Mixture Model(GMM)and Gaussian Mixture Regression algorithm(GMR).According to the variance of teaching trajectory,the gradient information is used to optimize the safety constraints of the reproduced trajectory.The motion reproduction is realized through variable stiffness control,so that the robot can maintain active compliant ability to the environment.Then,in order to ensure the safety and flexibility of compliant assembly control,a data-efficient learning variable impedance control method(ELVIC)based on reinforcement learning is proposed to solve the force constraint control problem.The Gaussian process model is used as the faithful dynamic model of the system.The utilization efficiency of the sampled data is improved by probability reasoning and planning,which makes the robot learn variable impedance control strategy efficiently.Experiments on a six-degree-of-freedom industrial manipulator show that the learned impedance control strategy integrates the advantages of high rigidity and flexibility.The effect of force control is not only better than that of adaptive impedance control methods,but also ten times faster than other learning variable impedance control methods.Finally,in order to verify the key technologies of robotic compliant assembly control system for explosive components,the compliant assembly control system is constructed by using Kinova lightweight manipulator.The assembly experiments of explosive components are carried out.The experimental results show that the variance information of the teaching trajectory can be effectively used to obtain the variable impedance control strategy which satisfies the safety constraints,achieving the integration of high rigidity and compliance.The ability to resist external interference of system is improved,which can effectively suppress the human interference.It effectively avoids the state of wedging and jamming,ensuring the safety and flexibility of the assembly process.It provides an effective control algorithm and technical path for the safe,efficient and intelligent assembly of explosive components.