Some good advice, thanks! But my lecturer thinks I should focus on testing the accuracy/calibration of the laser scanner by creating some tests. One idea he had was to test the maximum distance of the laser scanner and test what effects atmospheric conditions might have on the accuracy. Any ideas how i could do this?
Off the top of my head:
Scan some permanent faces from the same control points in sun and rain and fog and moon. Make some kind of interesting display that highlights the differences in accuracy. Do it at the edge of the scanner's published ranges. Maybe even make some kind of "contour map" where the point clouds of the same face have differences.
I myself would be curious to hear how much more practical range you could get at night vs. under full daylight.
Would you recommend setting targets at the maximum end of the laser scanners measurement distance, or scanning, for example say a building at the maximum distance (about 150m)?
Seems like you would want control near and far, and then resection your scanner occupations from those for the best position & oritentation. Maybe six control points, half the range out on the left and right, half the distance toward the building faces, and then some at the building face distance. You can compare the quality of the far control to the near control. Maybe you even want a 2nd building face that is nearer, so you can compare the quality of the scans near and far in different conditions. If you resectioned a bunch of positions at different distances from your building faces & your far targets, you could make a pretty graph of accuracy vs. distance in different conditions. I bet it's kinda linear to a certain point and then scatters.
Huh, So now I am back to your question, buildings or targets? You could streamline the first try by just using targets. You would then have to explore the difference between targets and scannable surfaces.
Yes, I am just making this up. No, I don't know what you will discover if you do it. That's what makes it interesting!
A study of how modeling a point cloud actually makes the data less accurate/precise.
hmmm if it doesnt have to be too survey based it would be a cool and quick idea to get different machinery or cars and set up the scanner in the drivers position to prove the machines/cars blind spots based on average height and such.
but im guessing this might be a little bit 'simple' or what you want.