LiDAR to Design to Survey Layout – Riegl VZ600i

[pfirmst]

Great video, I really like the part where you discuss accuracy. This is a common problem with laser scanner manufacturers, there just isn’t the same level of accuracy analysis as there is in surveying with least squares, nor is there sufficient information about the accuracy of the instrument itself.

This is a very fast scanner and that’s very impressive, but there is always an accuracy v’s speed trade-off.

[OleManRiver]

Nice job. It takes some due diligence to truly get to manufacture specs on using a scanner. After performing numerous scans in many different ways and testing against metrology measurements equipment if not done properly then almost in any scenario it’s a crap shoot. For accuracy. Same for lidar drones etc. at least you are proactive and testing so you know what and when it’s a good tool to use. Keep up the good work.

[pfirmst]

I don’t believe 5mm @ 95% confidence is possible for that scanner, but likely they’re quoting std deviation, which is 68% confidence, so 10mm @ 95% confidence.

It takes us 18 weeks of static observations to get down to 1mm @95% confidence, relative to fiducial network CORS reference stations.

Has any one else made the observation that GPS accuracy is improved when multiple stations being observed concurrently and are included in least squares analysis with other instruments, than with a single GPS station?

1 week of static observations for two positions gets us 5mm @ 95% confidence, but only for those positions where static observations are performed, we need to add 3mm for other positions, ie 8mm @ 95% confidence (I’m ignoring the coordinate with the poor reliability rectangle in the attached image as it was only measured from one position). For less than 24 hours static OBS, salsa reported 10mm @ 95% confidence.

If we say, relative to ground datums, it’s 3mm @ 95%, but we still need to add something for point cloud observation, points that aren’t least squares adjusted, but lets for argument sake allow 2 to 3 mm @95% for that. Remember that direct readings aren’t as accurate as prism observations, direct readings are affected by angle of incidence and intensity, so it’s likely worse than that.

Vertical is very tight, we used a first order level (overkill), for the scanner we’ve used a Topcon GLS-2000, fresh back from calibration, no total station was used, just the scanner, level and GPS base stations with prism targets, we’ve levelled relative to the prism occupation point where we didn’t mark ground points.

This is just a demo around a house.

If we capture 18 weeks of GPS, then we will likely have a point cloud 5mm in 3D @ 95% confidence, but we would need to add some observations from a 1 arcsecond total station to do it, with some allowance for point cloud observations, but that’s not really practical for small jobs.

So what happens; companies have scanner technicians (not surveyors), they purchase a scanner, quote the manufacturer’s accuracy to clients, then their clients engineers believe it and write it into their company standards, but they don’t allow sufficient time in their schedules to complete the work to the level of accuracy required.

Don’t get me wrong, it’s a great scanner, but it makes it hard when manufacturers quote levels of accuracy that aren’t achievable in real world conditions.