labs
Lab manual overview
How the labs are organised, what order to do them in, and where the criteria come from.
Status: the lab pack has not shipped. What follows is the first few that are planned, not the full set — labs keep being added as machine types are added, and the final list is whatever ships.
The labs themselves — goal, steps, pass criteria, common mistakes, review questions — live on the labs index. This page is about sequence and intent.
Order
| # | Lab | Teaches | Time |
|---|---|---|---|
| 1 | Pixel size | The first chain; residual ≠ deviation | 30 min |
| 2 | Hand-eye | Wrong configuration raises no error; closed-loop checks | 45 min |
| 3 | Motion error and backlash | Unidirectional vs bidirectional; repeatability ≠ accuracy | 45 min |
| 4 | Z focus | Sharpness curves; how noise moves the focus | 40 min |
| 5 | Galvo field | Grid density; why RMS alone hides the corners | 60 min |
Why this order
It is not sorted by difficulty but by what each one has to teach next:
- Lab 1 gets you through a complete chain and a real deviation number, and shows for the first time that a residual can lie;
- Lab 2 pushes "errors that raise no error" to its limit — the wrong configuration still produces a small residual;
- Lab 3 moves from vision to motion, so that error stops being something only the camera does;
- Lab 4 brings noise into measurement, preparing the ground for error budgets;
- Lab 5 combines everything, depends on earlier chains, and is closest to real project complexity.
For instructors
- Every lab contains a step that breaks something on purpose. Do not skip it. Producing the error yourself beats ten explanations.
- Labs 1 and 3 suit a first week; Lab 5 works as a final assignment.
- The numbers in the pass criteria can be changed for your own machine — in machine configuration, not in code.
Last updated: Sep 23, 2026
Was this page helpful?