Reflectorless total station self-calibration

Abstract

A method for reflectorless total station self-calibration of face left and face right observations in a network is proposed. The mathematical models for the observation equations augmented with systematic error terms are presented. The method is demonstrated on three datasets captured with two reflectorless total station instruments: a Trimble 5603 DR200+ and a Leica TCRA 1103+. The results show improvement of up to 71% in the RMS of estimated residuals as a result of self-calibration. Furthermore, most of the error model parameters can be estimated with sub-millimetre or sub-second precision. Statistically significant differences between some calibration parameters from two Leica datasets were found, but closer analysis shows that these are of no metric consequence. Incidence angle is demonstrated to be a significant factor in distance precision, as has been reported elsewhere for terrestrial laser scanners. Comparison of the results with those from independent component testing shows that the self-calibration method is superior for determining angular error parameters. A close agreement of 1.1 mm was achieved between the additive constant estimates from self-calibration and baseline calibration methods for the Leica instrument. The additive constant estimates for the Trimble instrument differed considerably, which is likely due to the observed non-linear behaviour of the distance meter at close range.