Kinematic model identification of planar delta manipulator using Random Levenberg-Marquardt algorithm

An accurate kinematic model is one of the fundamental requirements of high performance robotic manipulators. But due to the existence of tolerance and clearance, calibration should be always carried out first of all to identify the inaccuracy and further modification of the kinematic model. In this paper, the calibration work of a planar parallel manipulator was presented in details. And an innovative algorithm to solve the nonlinear optimization problem was proposed, which's named “Random-Levenberg-Marquardt” algorithm. With specific illustration and practical experiment result, this algorithm was proved to be efficient, accurate and easily implementable comparing to the traditional methods.