Multilevel Collaborative 3D Measurement Technology Of Intelligent Measuring Robot

With the rapid development in China’s economic and military fields,the measurement of irregular components in large equipment(such as aircraft,large ships and space stations)has become a hot issue in the field of instrument science.For the measurement of irregular components,the most basic method is the traditional measurement method,which is realized by the operation of various measuring instruments by professional technicians.It is difficult to get rid of manual intervention and has low efficiency.Therefore,a more intelligent and automatic measurement method is needed.In recent years,industrial robots with high-precision sensors have become a new way to measure irregular components,making the measurement more intelligent and efficient.This measurement method relies on the flexibility and programmability of the robot to automatically complete the measurement of the measured object.However,due to the inaccurate positioning of the end of the robot,there is a large error when the local measurement data of large components is converted into the overall measurement data,resulting in inaccurate measurement results.In this paper,a multi-level collaborative measurement technique based on the idea that first rough measurement and then fine measurement is proposed for component surface.The first-level measurement was completed by three Kinect depth cameras:firstly,the three Kinect depth cameras were calibrated twice by hand and eye,and the point cloud was transformed to the world coordinate system.Then,the scanning path of the end of the manipulator is obtained through point cloud slicing and normal vector extraction,so as to ensure that there is no collision between the manipulator and the measured object and that the measured object is always within the range of the on-line structured light sensor.The secondary measurement is completed by the line structured light sensor: after the hand-eye calibration is completed by the line structured light sensor,it is driven by the manipulator arm to scan along the above path.The three-stage measurement is accomplished by a laser tracker: the laser tracker is used to obtain the accurate position of the manipulator end in real time.Finally,a data synchronous acquisition program is written to receive and process the two-dimensional measurement data of the sensor,the pose data of the manipulator end and the coordinate data of the laser tracker in a synchronous manner,so as to obtain the three-dimensional point cloud information of the measured object,and complete the reconstruction of the point cloud.The results obtained are compared with the original model.In this paper,a measurement system based on UR5 6-DOF robot,linear structured light sensor,Kinect depth camera and laser tracker is proposed.In the Linux system,C++ is used as the programming language to write the trajectory planning program and the data synchronization receiving program.Combined with the Moveit robot planning library,Eigen matrix library and PCL point cloud library,the robot scanning path point extraction,robot trajectory planning,the filling and processing of point cloud data are realized.Finally,the 3D measurement experiment is completed.With the help of the original model of the measured object and the reconstructed point cloud after scanning,the registration error is less than 0.20 mm,which indicates that the measurement system has a high measurement accuracy.The reliability of the system in practical engineering is proved.