| In robotic surface treatment processes,force control has a significant impact on the quality of the workpiece.Unstable machining contact forces can lead to overcutting and undercutting of the end tool,which in turn directly affects the quality of the product profile and the machining accuracy.Therefore,this paper designs a surface treatment system based on a five-degree-of-freedom robot in the context of the surface treatment process,followed by research around a three-dimensional force detection platform,variable impedance control based on deep reinforcement learning and robot surface treatment experiments.Details of the research are as follows:(1)Design of a robotic surface treatment system.Firstly,A robotic machining system suitable for surface treatment has been designed based on the reachable space of a gantry-type five-degreeof-freedom industrial robot.At the same time,ANSYS is used to complete the static analysis of the end-tool.Under a pressure of 100 N,the maximum strain is about 0.228 mm and the maximum stress is about 10.81 MPa,which satisfies the industrial requirements of deformation and the maximum yield strength of the material.Then,the analysis of the robot’s forward and inverse kinematics is completed according to the MD-H parametric method,and the simulation is done in conjunction with the Simscape module.Finally,the robot dynamics model is established by using the Newton Euler method,and on this basis the design of the calculation torque control algorithm is completed.(2)To address the problems of high cost and poor universality of multi-dimensional force sensors,a design solution for a three-dimensional contact force detection platform is proposed.Firstly,the contact force of the surface treatment process is analyzed,and an XY force detection platform based on the slide rails is designed accordingly;secondly,in order to improve the accuracy of the force measurement on the platform,in addition to the coordinate transformation and gravity compensation of the Z-axis tool,the signal filtering and calibration processing of the force sensor are also realized,and the calibration results show that the maximum repeatability index of the force sensor is about 0.4%,and the non-linearity index is about 0.342%,which meets national standard;finally,the verification of the XY platform force measurement is completed by using the digital display tensiometer,and the errors of the XY two axes force measurement are 0.2472% and 0.2772%,which meet the actual use requirements.(3)In order to address the problems of conventional impedance control where the optimal parameters are not easily determined and the algorithm is poorly adapted,a variable impedance control scheme based on the double delay depth deterministic strategy gradient(TD3)algorithm is proposed.Firstly,the impedance control model for robot surface treatment is established,and the influence of impedance parameters on the system performance is analyzed;secondly,a composite force control algorithm is designed by combining the TD3 algorithm with the impedance control algorithm.The training speed of the model-free algorithm is improved by limiting the range of action space,introducing OU exploration noise and designing reward functions;finally,three simulation experiments were designed to verify the effectiveness of the algorithm.The results show that the method has a faster adjustment time and an average error of approximately 1.1mm and 0.01 N in position and force tracking respectively.(4)To verify the feasibility of the above theory,the removal of paint from the surface of the workpiece is carried out using both flat and curved workpieces as the object of processing.The experiments show that under the variable impedance control algorithm,the average error of flat and curved workpieces is 0.2658 N and 0.4352 N respectively.After the point cloud inspection,the area of flat workpieces with acceptable tolerance is about 85.6% and the area of curved workpieces is about 67.08%,which can guarantee the flatness of workpiece surface to a certain extent. |