Browsing by Author "Pexman, Kate"

Now showing 1 - 1 of 1
  • Thumbnail Image
    ItemOpen Access
    New method for first-order network design applied to TLS self-calibration networks
    (Elsevier, 2021-07-01) Lichti, Derek D; Pexman, Kate; Tredoux, Wynand
    Terrestrial laser scanning (TLS) is established as a viable means for precision measurement and the need for systematic error modelling and instrument self-calibration is well recognized. While additional parameter (AP) models and procedures for their estimation from signalized target fields have been developed, the first-order design (FOD) of TLS self-calibration networks remains an active area of research aiming to improve AP quality. The conventional FOD approach of numerical simulation carries a heavy computational burden. This paper reports a new method for TLS self-calibration FOD that avoids the high computational effort and can predict AP precision in closed form. Its basis is a relatively simple analytical model of the distribution of spherical coordinate observations, specifically the elevation angle. The accuracy of predicted AP precision is quantified by comparison of precision estimates from a more complex and detailed observation distribution model and from self-calibration. Results from 25 datasets demonstrate the high accuracy (arc second or better) of the closed-form approach. A new observation distribution model is then developed to optimize the geometric design of TLS self-calibration networks. An ideal observation distribution based on the versine function and a corresponding target field configuration that enhance AP precision are established. Testing was performed on five additional, very dense TLS self-calibration datasets. Each dataset was subsampled so as to replicate the observation distributions corresponding to conventional network design and the proposed design. The results show that up to 55% improvement in AP precision, obtained from self-calibration, can be achieved with the new design and these results agree with versine-distribution model predictions within 14% to 16%.