Figure 8: Error by Distance from Anchor (m)
However, as previously stated, some of the observed errors could be caused by the user's
ability to consistently anchor the model in the same location and with the same orientation. To
mitigate user error influencing these observations, a baseline analysis offers a suitable
alternative in which the distances are unrelated to the user's ability to precisely anchor the
model round over round. See Tables 8.1 & 8.2 in Appendix B for the full set of baseline
distances, observations, and errors. By assessing these baseline errors, a more realistic
assessment of the drift error can be gathered. These errors are derived through the following
formula: where represents the scale error found in section 1 above d ) e
T
= (
A−i
*
e
S
− d
obs
e
S
(-0.03%). The table below highlights the average baseline error in both WLT and non-WLT
solutions. Most notably, when utilizing WLT the average error is reduced by nearly 4x, providing
a more reliable and stable user experience. These results mirror the above positional errors and
confirm the conclusions drawn from Table 7.
WLT Integrated WLT Not Integrated
Statistic Error (m) % by Distance Error (m) % by Distance
Average 0.076 0.31% 0.303 1.27%
Median 0.051 0.28% 0.288 1.33%
Max 0.195 0.64% 0.689 1.62%
Min 0.005 0.10% 0.034 0.67%
Table 9: Statistical Analysis of Baseline Error (m)
13
