Research On Trajectory Optimization And Control Of Viscous Rubber Unstacking Robot

In view of the performance requirements of industrial robots in the special application for strong adhesive rubber unstacking,this subject has carried out research on rubber unstacking robot trajectory optimization and control.In this paper,the viscoelastic mechanical model of robot-rubber contact process is established.Based on this model,the optimal trajectory planning and adaptive backstepping control of the robot are studied under viscoelastic contact process.Due to the unknown adhesive force between rubber blocks,the research on robot improved impedance control strategy based on reinforcement learning algorithm is carried out under rubber separation stage.In addition,based on the industrial computer and high-speed communication bus technology,the experimental verification platform design of the rubber unstacking robot is completed and the corresponding performance test and experimental research are carried out.In order to solve the problem of dynamics modeling of robot inserted rubber process,a method based on Hunt-Crossley foundation model is proposed to establish the viscoelastic model,and the experiment system is designed to analyze the accuracy of the model.Firstly,the linear and nonlinear viscoelastic contact models are analyzed.By comparing the hysteresis loop and the power flow curve,the Hunt-Crossley nonlinear model is selected as the basic model.Based on the basic model,the mechanical analysis of the inserting process is carried out to establish the mechanical model,and the linearization process reduces the difficulty of solving the model.The Levenberg-Marquardt nonlinear least squares method is used to identify the parameters and reconstruct the mechanical model.The Kruskal-Wallis nonparametric test method is used to analyze the experimental results.It is verified that the mechanical model is much more approximate to the measured results than other linear models.In or der to solve the problems of dynamic modeling errors and strong external disturbances caused by insertion of viscoelastic rubber,a multi-objective trajectory optimization based on floating transition points and adaptive backstepping control method based on recursive fuzzy wavelet neural network(RFWNN)is proposed.Firstly,combined with the viscoelastic dynamic model and the robot operation characteristics,seven design indexes including operating efficiency,running trajectory smoothness,energy consumption and normalized moment are proposed,which are regarded as objective functions and constraints condition respectively.On this basis,a multi-objective trajectory optimization model with three objective functions is established to optimize the floating transition point of B-spline trajectory.The optimal trajectory of the robot inserted into the viscoelastic rubber process is solved using non-dominated sorting genetic algorithm(NSGA-II).Based on the optimal trajectory,the traversal method is used to calculate the optimal energy consumption place point of the whole stack rubber.Then,in view of the anti-jamming trajectory tracking control problem of robot insertion viscoelastic rubber process,the RFWNN method is used to deal with the time-varying uncertainty of the robot dynamics system,at the same time,an adaptive backstepping control law is designed based on the Lyapunov theorem to eliminate the dynamic estimation error and unknown disturbance.Finally,the simulation and comparison of the above control method has been carried out.It has been verified that the control strategy proposed in this paper can make the rubber unstacking robot complete the tasks exactly and smoothly according to the optimal trajectory.In order to improve the stability of the contact process between robot and unstructured environment,the improved impedance control is used to design the rubber unstacking robot control strategy due to the time-varying and unknown adhesive forces between the rubber blocks in the rubber separation stage.Firstly,the impedance control method based on dynamics and position is analyzed respectively.It is proved that the reason of traditional impedance control method can not simultaneously guarantee the accuracy and robustness of the robot impedance control system in the interference environment.Accordingly,an inner/outer loop impedance control strategy based on time delay estimation and NAC reinforcement learning is proposed,and the stability conditions of the control system is derived.The inner/outer loop control method is used to apply the required impedance and correct the modeling error caused by the unknown disturbance of the adhesive force between the rubber.Besides,the time delay estimation method is used to estimate and compensate the nonlinear dynamic term of the robot.And a recursive least-square filter-based natural actor-critic reinforcement learning algorithm is used to optimize the impedance parameters of the control system.Through the analysis,it is demonstrated that the above control strategy can improve the impedance accuracy and robustness of the robot control system under the unstructured dynamic environment.So that the robot can work smoothly under unknown adhesion force to reduce the vibration and improve the rubber separation performance.In experimental platform establishment of the rubber unstacking robot respect,a real-time robot motion control system based on industrial control computer(IPC)and high-speed communication bus techonology is designed,and the robot bottom control software is designed based on the motion control state machine.The servo cycle timing is completed using Twin CAT which is a windows-based real-time system.The servo control commands are passed to the drive via the Ether CAT bus during each servo cycle and feedback of the motion commands is completed at the same time.Finally,the viscoelastic contact experiments of the rubber unstacking robots inserted into the rubber process and the rubber separation experiments with unknown adhesion force were carried out to verify the effectiveness of the proposed control strategy.