THE IMPACT OF THE DISTRIBUTION OF OBSERVATIONS ON TERRESTRIAL LASER SCANNER SELF-CALIBRATION QUALITY

Abstract. Terrestrial laser scanning (TLS) instruments find routine use for a range of precision engineering measurement applications. Depending on the accuracy requirements for a specific project, the instrument may require self-calibration to determine systematic error model terms. One of the goals of first-order network design for self-calibration is to acquire observations throughout the full instrumental field-of-view. Experience calibrating TLS instruments has demonstrated that while this goal can be achieved for horizontal deflection angle observations, it is seldom realized for the vertical angle observations. This paper presents results from a preliminary investigation into the impact of the distribution of vertical angle observations on the estimation of two critical systematic error parameters in TLS instruments: the collimation axis error and the trunnion axis error. First, a model to characterize the empirical observation distributions is developed. The model is a function of a single shape parameter that quantifies observation dispersion. Then, a means to estimate the impact of the distribution on the parameter estimation is developed. Results from six real datasets show the distribution model characterizes the overall general trend of the observations. Simulated results show the relative independence of the collimation axis error and the strong dependence of the trunnion axis error on the shape parameter.