AtlasMeter Testing Continued

I fixed the lost steps issue by taking inspiration from my CNC machine. Since they don't have infinite acceleration, CNC machines use trapezoidal velocity profiles to slowly ramp in and out of movements. My camera dolly also can't accelerate infinitely, so I applied a trapezoidal profile to the velocity of every camera move. After a bit of tweaking, I was no longer losing significant number of steps. I tested from 10 to 204 frames and every time the dolly moved the right distance. However, the dolly now travels a little slower (the move step function now takes a delay parameter and therefore takes more CPU time). I need to implement variable speed movement to save time in setting up the start and end positions of the dolly.

AtlasMeter Testing

After running the timelapse dolly several times, I've uncovered several bugs in the code and hardware (and fixed them). I added a screw to mount the camera (without the mount the camera slowly shifted around) and fixed some time display bugs. For some reason, the dolly consistently misses steps, and the more frames there are the more steps are dropped. At the end of a 240 frame run that was supposed to travel over the entire length of the dolly the camera was nearly a foot short of the set end position. I'll have to fiddle around and add some constant to the steps per frame to get the right distance.

I noticed that the camera moved slightly faster in the beginning than at the end and didn't have a very consistent speed  in the middle. There must be more steps dropping or some play in the belt. Also, I need to remember to not move the camera too much per frame or the video is choppy.

AtlasMove Rev.0 Complete

My timelapse dolly project, named AtlasMove, has finally become a functional prototype. I just finished programming a rudimentary menu system for setting up the timelapse parameters. I still need create clamps to hold the belt at each end of the slider to keep the belt taught.

Front side of the circuit board

Back side (a little messy)