HRD Based Robot Programming And Force Control For Robotic Manufacturing Of Complex Structures

The processing quality of the blades directly affects the performance and lifespan of the turbine engines.At present,the application of automatic polishing technology for the known surface of the blade is well-rounded.However,for complicated structures such as the blade edge and the damper table,due to the difficulties in programming and configuration,manual polishing is still the most widely used method.However,the efficiency is relatively low and the consistency cannot be guaranteed using this method.In this thesis,a robot demonstration-assisted programming-based force-controlled robotic polishing method is proposed for complex structured blades.Two essential technologies,namely the compliant human-robot interaction and force control of robot-environment contact were studied.The specific research contents of this thesis are as follows:1.Robot control model construction.Kinematics and dynamics analysis were carried out for the serial robot,which is commonly used in industry.The kinematics and dynamics models were established in Matlab environment.By analyzing the motion control and force control frameworks,it can be concluded that motion control can guarantee the accuracy of force control.Thus in order to develop the accuracy of force control,a force control framework with inner motion control loop is proposed.The simulation results show the effectiveness of the proposed method.2.Design and improvement of adaptive motion controller.For the inner loop motion controller of the proposed compliant control strategy,firstly the driving torque model is established in Cartesian space.The proportional derivative(PD)tracking controller is adopted to realize the trajectory tracking control.The robust adaptive PD controller is adopted to improve the tracking accuracy and robustness.Finally,a radial basis function neural network adaptive control method is proposed to further improve the tracking accuracy,and there is no need to linearize the robot dynamics model using this method;3.Human intention prediction-based compliant human-robot interaction.Aiming at the compliance during the human-robot interaction process,firstly the linear quadratic regulator model of the human-robot interactive system is established,and the Integral Reinforcement Learning method is used to optimize the impedance parameters,which can reduce the operation force and improve the compliance.In order to eliminate the delay which is caused by the impedance model,a force feedforward compensation method is proposed,which can further reduce the operating force and improve the trajectory tracking accuracy.4.Constant force tracking for complex environment.Aiming at the problem of constant force control during the interaction between robot and environment,firstly a reinforcement learning based constant force tracking method is proposed.After multiple offline trainings,accurate force tracking can be realized.Then a constant force tracking method based on reference trajectory online tuning is proposed and improved,thus the online constant force tracking control is realized.From the simulation experiment,the force tracking accuracy is improved by 30% before improvement.Finally,experiments on the apparatus verify the feasibility and effectiveness of the proposed method.