The development of intelligent manufacturing has promoted the development of robots.At the back end of the manufacturing industry,intelligent three-dimensional detection of products has become increasingly important.With the development of robot technology and visual measurement technology,the technology of using robots equipped with visual sensors to measure objects in three dimensions has been extensively studied.However,the traditional three-dimensional vision method mainly relies on the teaching method of the workers,which relies more on the experience of the workers,and the detection method is difficult to standardize.Therefore,it is of great significance to study a measurement system that can intelligently plan the scanning path of the robot according to the shape of the object without relying on the prior three-dimensional model of the object to be measured.Based on the above background,this paper proposes a two-level collaborative measurement method.First,multi-Kinect collects the object point cloud for first-level rough measurement,and then obtains the scanning path of the robotic arm.The key issues such as measurement model establishment,multi-kinect point cloud registration,scanning path planning,and robotic arm motion planning were studied.The following is the main research content.Established the forward kinematics model of the six-degree-of-freedom manipulator and the measurement model of the line structured light sensor,and established the measurement model of the three-dimensional measurement system;Kinect has a wide measurement range but the accuracy does not reach the final measurement accuracy.The angle of view of a single Kinect cannot be limited Collect the complete three-dimensional information of the object,so use three Kinects arranged around the work space to get a three-point work space point cloud and then register it using the ICP algorithm,according to the size of the object to be measured and the sensor measurement range.Sharding,determine the scanning path of the robotic arm according to the slicing,calculate the normal vector of the point cloud on the scanning path,and determine the mechanism according to the height of the sensor and the measured object and the attitude of the sensor relative to the measured object during the measurement of the line structure light sensor Scanning position of the arm.Based on the NURBS curve to describe the good performance of the free curve,the NURBS curve interpolation algorithm is used to interpolate the position of thediscrete manipulator;based on the quaternion to describe the uniqueness of the rotation and the linear nature of the interpolation,the quaternion spherical linear interpolation is used to machine The posture of the arm is interpolated;the smooth trajectory of the mechanical arm is obtained.UR5 collaborative robot and line structured light sensor are used to establish the hardware platform of the measurement system;the ROS software platform is used to write a synchronous acquisition program of sensor data and mechanical arm joint angle data,and a real-time display program of the three-dimensional point cloud of sensor measurement,which realizes the multi-kinect depth camera The function of the scanning path of the robotic arm is obtained through measurement.The position and pose of the manipulator are interpolated and the motion planning of the manipulator scanning is simulated in the ROS system,and the effectiveness of the interpolation algorithm is verified. |