Observability index optimization of robot calibration based on multiple identification spaces

A calibration method is proposed for six-DoF serial robot based on multiple identification spaces consisting of two subspaces in which the orientations of joint 3 and poses of end-effector are measured simultaneously using hybrid sensors. The rotational geometric errors with higher sensitivities are identified in the first space while the rest are identified in the second. Compared with single identification space used in traditional methods, the number of geometric errors to be identified is reduced in each subspace. Thus the identification vectors corresponding to the geometric errors belonging to identification models can be better spaced. Simulation results show that the observability indices and identifiability are further improved by using the multiple identification spaces. Experimental results are also obtained from a six-DoF serial robot with laser tracker and IMUs to verify the identification accuracy improvement. Uncertainty analysis of each identification results is also provided.