Study On Model Error And Calibration Method Of The Measuring Robot

The general measuring robot system based on stereo vision is one of the significant engineering fields of the vision measuring technology applications. The system comprises an industrial robot and a stereo vision sensor mainly. So, it not only keeps the vision measuring technology's characteristics of non-touch, quickness, but also extends the working space of the sensor and enhances the flexibleness. The stereo vision sensor collects the image signals of the given point at a certain posture of robot and then obtains the three dimension coordinate data after computing the image data. In order to ensure the measurement accuracy of the system, every part should be calibrated. The most difficult step is the calibration of the robot, i.e. establishing the correct mapping from the joint variables to the position and rotation of the end-effector when robot is at any posture. The main contents are as follows:1. Real-time communication between computer and robot was realized. In addition, the robot was programmed by using RAPID language so that it could change postures as schedules and record relevant data at calibration postures in text file.2. The robot was modeled based on the D-H convention. The forward, inverse and differential kinematics problem was solved. Then, simulative tests for each process were taken in manner of the coupling of Matlab and VC.3. A linear solution method of acquiring the real link parameters was presented. Adopting CPC model instead of D-H model, parameters'singularity and jump caused by model flaw was eliminated. The calibration process was divided into two steps in which orientation and translation parameters were solved separately. The orientation parameters were solved for recursively and then they were put together into the translation equation to solve all the remaining translation parameters simultaneously.4. The algorithm for the abovementioned method was studied. Sequentially, the robot calibration routine was programmed by using Matlab and was debugged.5. The whole calibration experiment was accomplished, including data collection and processing, etc. Calibration result was gained. Some conclusions were put forward by comparing the errors before and after the calibration.