Measure the real arm

Identify a servo you can’t open, with error bars on everything.

Identify the servo on the pendulum bench and the torque stand, measure the printed arm’s masses, delays and noise, and check that every measured interval lies inside the ranges the model was trained on.

Written lesson + build labEvidence-based gate

The goal for this build

What you’ll be able to do.

Identify friction, backlash, the torque-speed limit and the gain mapping of a servo whose firmware you can’t read.
Measure torque, current, masses, inertias, delays and channel noise with an instrument that can observe each.
Decide from intervals, not means, whether the trained model has seen physics like the real arm’s.
Put the melody on the beat with the delay your servo model predicts.

What to have ready

Module 7 gate passed, including the camera noise from the fixed tag
The pendulum bench and the torque stand on the sixth servo, with the bench Pico 2 streaming the load cell
The Module 5 randomization ranges, centered on the CAD parameters since Module 6
The Module 7 melody recording and the first wave’s servo logs

Download the course starter ↓

Free account

Read the lesson, run the lab, pass the gate.

The whole course is free. Create an account to unlock the lesson, the build lab, the task and the gate command for this module, and to download the starter kit.

Create a free account Sign in

03 / Prove it

Keep the evidence.
Then run the gate.

What you ship

params_real.yaml with units, error bars, methods and raw-data references
A CAD-versus-real mismatch table, with each interval set against its Module 5 range
The identified servo model with its held-out pendulum runs and payload currents, and BAM’s fit of the same data beside it
The wave’s sim and real joint traces overlaid, and a bench.py sim rerun with the measured values
The melody recorded on the beat, with its timing compared with Module 7’s

Module 08 gateEvidence required

Every parameter is measured with error bars, or declared unresolved with an assumed range from a named source. Link masses are always measured.
Each interval or assumed range lies inside the Module 5 randomization range; where one doesn’t, that range is widened and the model retrained before Module 9.
The identified servo model predicts held-out pendulum runs within 1° RMS.
The currents measured while holding three known payloads match the model within 10%.

With a free account: the gate command and how to record your evidence.

The story behind the build

Episode companions.

EP 09How wrong was my CAD? Identifying a servo I can’t open

Written companions are available now. Recordings and customer access are forthcoming.

Made it through the lesson?

Save your reading progress here. Record your gate result in your project.