Research On Attitude Detection And Attitude Error Compensation Of Collaborative Robot

Attitude detection of collaborative robot can help the robot to identify the environment,understand the task.That can improve the safety,accuracy and efficiency of the robot.For robots with unkown configuration parameters,attitude detection can estimate their parameters.It's avail for further kinematics and dynamics analysis.For the economical collaborative robots,their kinematics parameters can be calibrated via attitude detection.A dependent robot attitude detection system can be used as a secondary auxiliary system,together with the high-precision encoder system of robots,to form the robot fault diagnosis system.This paper studies the attitude detection technologies of serial collaborative robots.The main research contents and results are listed as follows.First of all,this paper studies the research stastus of attitude detection technology at home and abroad.After compapering different schemes,the inertial-magnetic-based detection technology is finally determined for attitude detection of collaborative robot.Secondly,this paper studies and achieves four common data fusion algorithms.Taking the raw data of the accelerometer,magnetometer and gyroscope,read from the inertial-magnetic measurement unit,as input,this paper realizes four attitude estimation algorithms based on the four methods.The performance of these four attitude estimation algorithms on the accuracy and computational burden are assessed via specific experiments.Thirdly,aiming at the errors existed in the attitude detection algorithms,this paper proposes a compensation algorithm for the attitude estimation errors based on the back propagation neural network.By importing the third party system as the reference,the model between the raw data of sensors and the attitude estimation errors is established based on the strong non-linear reflection ability of the neural network.The attitude is predicted by taking the raw data of sensors as inputs.Then it is used to compensate for the attitude estimation of the inertial-magnetic measurement unit,which improve the accuracy.The public data is utilized for the experiment,and the result shows that the proposed method can effectively improve the attitude estimation results of the inertial-magnetic measurement unit.Fourthly,aiming at the serial manipulator,this paper proposes a universal joint-angle estimation algorithm,using the inertial-magnetic measurement unit.In this method,the inertial-magnetic measurement unit is mounted on each joint of the manipulator.By decomposing the compound rotation of the joints,the attitude describes single rotation of each joint can be obtained.After calculation,the joint angles can be estimated.The Aubo i5 manipulator is used for experimental setup to validate the performance of the proposed method.Results show that the proposed method can estimate the attitude of the joint angles effectively.Finally,the client software is developed based on the windows presentation foundation.It includes the functions of robot control,communication,real-time data display,real-time attitude display of the robot's 3D model,data saving,etc.The attitude of the serial manipulator can be estimated in real time.