William A. LeonardComputer Consultant3D ScannerSummaryThe scanner project, so far, has been less than successful. I knew from the start that the design was inefficient because the scanner would have to "hunt" for the convergence - by trial and error moving forward and back until the two laser beams crossed at the target point. The problem was made worse because the still image from the web cam took about a full second to become ready. A coarse scan would take overnight to complete. The platform is heavy and there is some sag if the platform extends forward by any significant amount. It doesn't affect the laser projections but the camera is mounted to the Z-rails so it is affected by any sagging. The laser projection isn't always the brightest spot - even with the room lights out. That's the only explanation I have at this time. Once in a while, the displacement between the laser projections would be reported as excessive and the platform would move by a large step when, in reality, it had been converging nicely. As mentioned before, the platform is quite heavy and there are undoubtably some inertial problems so positioning may not be precise. The Z-axis movement is consistent forward and back but the X-axis movement is not consistent left and right. There is a discernable noise from that stepper motor when rotating clockwise. Does anyone know of a source for purchasing a gear puller? This was a challenge during construction when I changed motors but used original gears and belts from the printer mechanisms. One gear was damaged but with care, I was able to file down the burrs with a jeweler's file. There is probably some movement irregularity but the damage contributes little to the scanning errors... there are more serious problems to solve. I believe the project has reached a point where it needs a redesign. My thoughts at this moment tend towards eliminating the Z-axis movement (and the weight of the additional rails). For example, taking regular steps in the X direction, capture an image at each step. At the completion of the stepping, treat the images in the same manner as projections of parallel lines as in Thomas Tong. There are other successful devices using a variety of methods. My math may be made simpler because my light source would move linearly instead of angularly (as in other implementations) and the camera would have a constant position and orientation with respect to the light source. (Then again, maybe not.) This version of the project is a failure but, still, I had fun doing it and I'm a little proud that the electronics seem to have worked well. I'll try to keep these pages updated with the progress of the next iteration. |
|
11/22/06
Home
|