calibration + end to end test

[unglazedstamp]

• [x] calibrate the app coordinates
• [ ] try to use it end to end

[gruvw]

Status update:

We decided during the meeting last Tuesday (14.05) to finish the hardware "details" that were not ready before our team meeting on Thursday (16.05) at 11 AM. However, the different components were still not ready for end-to-end testing on Thursday, so we decided to meet again on Friday (17.05) at 1 PM. Some progress was made, but once again many components were not ready for calibration and end-to-end testing/demo of the full system with every component working.

Things that are required for calibration and end-to-end testing/demo:

• (not ready) Have the CV server ready with the edge detection or segmentation up and running (testing the CV automatic configuration)
• Be able to control the steppers from the new ESP board (move the head to correct position)
• Have a fixed camera module that does not have any play (for correct calibration using the markers and taking stable reproducible pictures)
• (not ready) Be able to control the servo from the new ESP board (for correct calibration and first button press)
• (not ready) Have a clamp that is fixed and works correctly (to have a correct point of reference for calibration)

Therefore, I was not able to do the calibration or "first button press" nor on Thursday or Friday.

Just tell me when all the requirements are ready (message me) and I will find a time slot to complete this task 👍

[gruvw]

So we tried the calibration again twice this week, and we are getting closer, but there are still a few problems that are holding us back:

• We are trying to find a way to precisely measure the corners of the markers on the camera picture (and display using simple python script to double-check), @nourguermazi working on it
• We are also running into some precision issues that could come from the steppers or the cart running into cables, @unglazedstamp working on it
• Meeting on time and getting things ready quickly

It would be nice to be notified when the requirements are ready.

[gruvw]

(see PR #55)

[gruvw]

Status update:

We tried again today, but we ran into issues with the belt that was too loose and started moving out of the way.

Therefore, I redesigned the belt holder, as you can see below:

I did two test prints on smaller parts just to try it out, and it was holding the belt tight in place. I updated the assembly with this new change (see onshape plotter assembly).

[gruvw]

As a positive, I ran the testing camera coordinates calibration script (that I wrote last week) on a picture of the real markers on the final clamp and it works wonderfully:

Left is in pixel coordinates, right is in motor steps after camera perspective correction.

[gruvw]

So yesterday, we achieved our first (actual) button press 🔥

See the result here: https://imgur.com/ZjUEPrZ

We will continue with the calibration of the CVserver and end-to-end tests today.

[gruvw]

So today we did our first integration test with the full system: controller + camera + server + application.

We found a few issues on the communication layer, but we managed to fix them and successfully configure a remote from the server, so full integration now works.

However, we are still lacking precision on the plotter... We are trying to fix this by changing the axes/gears that are in contact with the belt.

[gruvw]

We successfully performed a full calibration and end-to-end test today. The camera coordinates mapping worked on first try after the calibration.

The whole system works pretty well, most of the time. The precision issues and now fixed: by going slower and by changing the pins in the axes (previous message).

However, we are currently still running into some issues but with powering now. The whole system is supposedly under some current spikes sometime, and @unglazedstamp is working on it with us right now.

Anyway, I will close this as completed as the calibration and end to end integration tests are fully operational.