Kinematic Calibration of a 2D rotation platform

As known, industrial manipulators contain kinematic errors due to inaccuracy in manufacturing and assembly. When a robotic workcell is considered, other sources of errors arise because of inaccuracies in the relative location of different robotic devices. As a consequence, the repeatability and accuracy of the whole system are affected. Even though nowadays most of the robots can guarantee sufficient repeatability and accuracy, sometimes it can be appropriate to compensate the just mentioned errors by modifying the actuators motion by means of special algorithms and experimental measurements. In the particular case hereinafter described the goal is represented by the improvement in accuracy of a 2-DoF rotation platform [1]. This platform is employed inside a micro-assembly robotic workcell together with a 4-DoF manipulator with Schönflies motion and a vision system. Specific calibration procedures have already been applied to the robot and the vision system [2] and it is necessary to calibrate the platform in order to reduce the effects of internal errors on its positioning with respect to the other components. To perform the calibration task it is necessary to have a measuring system able to read the pose of the manipulator in different configurations, a mathematical model to correlate the unknown geometrical parameters of the platform with the pose of the end-effector and an algorithm to estimate the same parameters on the basis of experimental measurements [3]. The paper concerns all these aspects, from the realization and calibration of the measuring system [4] to the development of a model of the platform and the estimation of the parameters. The measuring system is based on a laser displacement sensor mounted on the manipulator gripper. When the laser beam hits an object the sensor measures its distance and, since the manipulator gripper pose is known, it is afterwards possible to calculate the spot absolute position. If the spot is moved on a sufficient number of points on to the platform end-effector, its pose can be measured. By repeating the measurements for a large number of configurations and performing the parameters identification procedure, the platform is eventually calibrated.