Calibration of parallel robots based on orientation constraint and total least squares approach

To improve the accuracy of parallel robots in an efficient and economical manner, a novel kinematic calibration method is proposed, based on the orientation constraint of keeping two attitude angles of the end-effector constant and the total least squares approach. The calibration is formulated as a nonlinear least squares minimization problem considering both deviations on kinematic parameters and measurement noise. The most likely combination of deviations is estimated by solving this optimization problem. Physical realization of the orientation constraint can be automatically achieved by levelling a low-cost biaxial inclinometer installed on the end-effector, rather than using any mechanical fixtures. Thus the proposed method has the advantages of no need for precise pose measurement, no special requirements on the actuators, and being independent of the measuring range and positioning accuracy of the employed inclinometer. Simulations show the effectiveness of the proposed method.