Research On Complex Noise Filtering Algorithm Of The Force Measurement Signal And The Compliant Control For Grinding Industrial Robots

If the contact force is not effectively controlled,the robot and workpieces may be damaged and the operation will not be completed successfully in grinding industrial robot contact operations in manufacturing,aerospace and other fields.As research has progressed,classical force control strategies such as impedance control and hybrid force/position control have been widely used in scenarios where information about the contact environment is known.However,in contact operations with uncertain environments,where the stiffness and position of the environment are difficult to know precisely,achieving effective compliant control is an urgent issue in contact-type operations.The research has important theoretical and engineering applications.In addition,in order to achieve accurate compliant control,the force sensor is typically required at the end of the robot to obtain contact force information,which often contains Gaussian white noise,vibration and colored noise superimposed to form a unique complex noise that requires special filtering.Therefore,this topic focuses on the problem of robot-perception filtering and compliant control under unknown environments.(1)Research on dual Kalman filtering algorithm for complex noiseTo address the problem that complex noise such as Gaussian white noise,vibration and colored noise are difficult to filter out effectively by conventional filtering methods in grinding conditions,a double Kalman filtering algorithm with a serial structure is proposed.The spectral characteristics of the force measurement signal during the grinding process are analyzed in detail and different treatments are carried out for Gaussian white noise and analysis of colored noise characteristic features: 1)The Gaussian white noise and highfrequency vibration spectral peak groups are filtered by the classical Kalman Filter(KF)algorithm;2)For colored noise filtering,the noise characteristics are analyzed.To descript the colored noise characteristics,a single parameter is introduced to approximate the variance of colored noise,so that an improved Kalman filtering algorithm is designed;3)The two filters are connected in series to form a double Kalman filter(DKF)algorithm,which effectively removes complex noise from contact-like operating conditions and extracting useful information from the force measurement signal.(2)Research on compliant impedance control strategy integrating feed-forward compensation and variable-damping modelTo address the problem of unknown or changing environmental position and stiffness in robot-environment contact operations,the position-based impedance control algorithm is difficult to effectively achieve compliant control,and the static and dynamic performance of the system needs to be further improved,the compliant impedance control strategy based on the fusion of feed-forward compensation and variable-damping model is proposed.The position-based impedance control strategy is studied separately from two aspects of considering steady-state error and dynamic performance: 1)The nonlinear integral PI is improved as feed-forward control and a nonlinear function with force error as the independent variable is designed to dynamically adjust integral speed to effectively compensate the reference position and reduce the steady-state error of force control;2)An activation function describing the relationship between force error and damping coefficient is designed,and a convolutional layer is added to extract useful information from the data so that a deep neural network variable-damping model can be built to adjust the damping coefficient in real time to suppress system overshoot and improve the dynamic performance of the system.The algorithm has been validated by simulation to be able to adapt to unknown environments and achieve effective force compliance control in different contact environments,whether constant force compliance control or dynamic force compliance control.A compliant control system platform based on the Yaskawa industrial robot MOTOMAN-GP7 is constructed.For complex noise problem in the force measurement signal,the experimental results show that the DKF algorithm is more effective than the KF algorithm in filtering out force measurement signal noise and has the features of low computational complexity and high practicality.And for the problem of compliant control under unknown environmental contact,the experimental results show that the proposed compliant impedance control strategy which integrates feed-forward compensation with variable-damping model,has better force tracking performance and effectively reduces the overshoot of force control.