Well I have made good progress on solving the issue of having straight lines show up as curved in the scan pointcloud. The cause is indeed caused by drag in the cables connecting the Sweep module to the PI Motor Hat. I first attempted to solve the problem by being more careful in routing the connecting cable and assembling the scanner turret at the 90 degree position. This is mid-way thru the scan and should balance the amount of movement of the cable. I also enlarged the length of the hole the cable passes thru. In future designs you should consider making this hole larger and with rounded edges. I also wound the Sweep multi-wire cable around the sweep module to shorten it enough so that the Serial/USB converter was up inside the turret. Unfortunately, all of this did not succeed in eliminating the drag/torque from the wires. Where I wound the Sweep cable around the Sweep module would shift and create a binding on the Sweeps rotation. I had solved the problem of the drag on the stepper motor with a new problem of creating drag on the Sweep module. The problem should be curable by taping the Sweep cable in place when it is wound around the Sweep module. But as I was dis-assembling and re-assembling the turret I noticed another simpler solution. By enlarging the hole in the side of the turret to slightly more than 10mm, the Sweep interface cable and USB cable can be moved outside where there is less flexture. In fact this seems to be working quite well as I can now make good scans without introducing a drag caused distortion. I can also say that having a larger hole to tighten the set-screw makes that chore much easier. I don't say that routing the cables this way is a good fix for the problem, but it is a workable solution. At least it allows me to further test the scanner.
All of this work re-assembling the turret unfortunately took its toll on the set-screw. Relying on a small thickness of plastic threads was too much to expect for repeated assembly. I ended up epoxying a small bolt and nut into the hole. I oiled up the threads on the bolt but not on the nut. That way the epoxy doesn't bond to the bolt threads but does to the nut. Once the epoxy started to set I began to turn the bolt back and forth a few turns to insure that it was free. The result is I now have a very solid set-screw.
And now, an actual scan. Attached is a scan I did outdoors in a nearby park. It features two large Saguaro cacti. I'm standing next to straight 3-4 meter tall one and also pointing at a 8 meter saguaro with multiple arms. The scan is done at 1Hz/1000pps and took about 9 minutes. VERY NICE!
I have one question about the scan. I see many, many extraneous points floating around in what should be open air. This is particularly noticeable if you zoom out a hundred meters or more. Do you know what the cause is? I assume that it is noise that the LidarLite is generating from the very high scan rate when the module is pointed to infinity with no target. I'm guessing that if I ran the scan at 500pps, I would have much fewer of these extraneous points. Something that I will be experimenting with.
Overall this is quite an amazing scanner, I'm very impressed with the quality.
I also have a suggestion on the Visualizer software 0.12v (perhaps added already in 0.14v?). Could you display the name of the data file being viewed?
Tyson, you mentioned that there had been issues with introducing a skew when the turret was not perpendicular to the stepper shaft. This results in scissor like 'cut' in the point cloud. I also refer to them as a screw-dislocation such as the similar defects in a crystal lattice. As you mention this can be corrected in software. I had similar problems with my previous scanner. I found that making a software correct was much easier if I added one additional scan swing to the whole scan. This extra scanline should be the exact duplicate of the first scanline except for the slight rotation introduced by the mis-alignment. This makes determining the rotation needed to correct the scan much easier and more robust.Saguaro331.csv (3.1 MB)