Compliance Control For Flexible Integrated Robotic Joint Based On Torque Estimation

Integrated flexible robotic joint is the core component of next generation industrial robot.The mechanical structure and controller design directly affect the performance of robot in actual working conditions.Unlike traditional robotic joints,different mechanical parts including motor,reducer,brake,sensors and corresponding drive and control circuit are integrated in one joint system.The advantage of the design is that the joint system can be repeatedly disassembled and installed,and easy to be maintained which makes it able to match any configurations according to different working conditions.However,in order to improve the output torque to self-weight ratio and realize compliance control,the harmonic drive and torque sensor were introduced into the joint system.The mechanical structure of harmonic drive is complex which makes it more difficult to model,meanwhile,harmonic drive and torque sensor bring flexibility into joint system and flexibility causes torque vibration.On the other hand,the torque sensor is expensive,which greatly improves the manufacturing cost of robot and limits the popularization and application of robot.Traditional robots focus on location and tracking tasks.In the past several years,human being have put forward higher requirements on robot.One of the basic requirements is to achieve friendly interaction between robots and environment as well as being safe to human.Therefore,compliance motion controller has received extensive attention in recent years.Traditional robots always focus on the requirements of rapidity,accuracy and stability.Compared with traditional views,compliance control also focus on safety and compliance with outer environment.Based on the above analysis,the following research is carried out in this paper:Firstly,based on the preliminary understanding of the joint structure design,a joint experimental test platform was set up.A servo control system for the joint system was designed,the controller parameters were tuned based on the dSPACE hardware on-loop simulation platform.At last,the available current,velocity and position controller was obtained and the above work give a foundation for the further researchSecondly,based on the mechanical structure design of integrated flexible robotic joint,the accuracy model system was established,a series of experiments were carried out to identify the model parameters.Nonlinear friction and dynamic gravity were compensated which have great influence on the motion control performance.Finally,the results of identification and compensation are verified by experiments.Thirdly,a torque estimation method based on neural network algorithm was proposed by analyzing the mechanical structure of the main transmission parts: harmonic drive.The input variables related to the output torque of harmonic drive was obtained to satisfy the mapping relationship.Finally,the difference between estimated torque and torque sensor information is compared by experiments,which verified the effectiveness and accuracy the estimation method.Fourth,in order to cope with torque ripple caused by the system flexibility,a full state feedback controller based on torque estimated is used to reduce the motor inertia and friction.The torque ripple was reduced and the tracking performance was improved.Similarly,experiments were taken to verify the controller.