Research On Torque Compensation And Detection For Collaborative Robots Based On Parameter Identification

Under the policy background of "Made in China 2025",it brings forward high request for intelligent manufacturing in various production fields,and collaborative robots develop quickly.Although the traditional collaborative robot can achieve the goal of man-robot collaborative control,the accuracy highly depends on external sensors,and the mode of position control is widely used to ensure the accuracy,which makes it difficult to realize the torque direct compensation control.Therefore,this paper studies the torque control algorithm of a kind of collaborative robot.Without external sensors,the collaborative function is realized through the algorithm.Collision detection and zero-force teaching control are realized through the complete dynamic model and the compensation of nonlinear friction model.Specifically,the DH coordinate system was firstly established for the collaborative robot,and then the dynamic model was established.Secondly,a dynamic identification model was constructed and linearized to determine the minimum parameter set.A multi-degree-of-freedom stepwise identification was proposed,and the excitation trajectory of the identification process was divided into two parts to improve the convergence speed of the trajectory parameters.Taking torques sampling noise as the weight,the weighted least square method was used to calculate the value of the minimum parameter set.The friction model adopted the cubic polynomial model,and the second-order Fourier series about the joint angles was added to the friction model to build a complete dynamic model considering the nonlinear friction.Finally,the overall error and the accuracy of the friction compensation were calculated according to the trajectory verification.The load identification was carried out by using the novel single joint identification method and the excitation trajectory identification method.The identification results were compared with the CAD measurement data of the 3D model.The results show that although the accuracy of single joint identification is a little low,the data collected is small and the identification speed is fast,which has a high application value.Based on the impedance control and the results of parameter identification,a torque compensation algorithm was established in the joint coordinate system to study the zero-force control.Based on EtherCAT Ethernet technology,the algorithm was constructed in TwinCAT development environment,and the front interface was developed in LABVIEW environment.The teaching force was measured by the dynamometer.After the teaching torque calculated,a comparison was carried out with the related algorithm.The repetition accuracy was verified by position collection.Collision detection was carried out on the basis of momentum deviation observer,and the collision torque observer was constructed by combining the first-order observer with friction compensation to achieve torque tracking.High frequency noise was filtered by a first-order low-pass filter.After that,a simulation was carried out to verify the model and optimize parameters,and an experiment was carried out to measure the collision force.After measurement,the maximum collision force was less than the human body's bearing limit,thus ensuring the safety of collaborative robot.