Part 02 / Hardware & flight

08 / 09

Measure the real drone

Replace convenient constants with measurements and error bars.

Identify thrust, inertia, actuator response and observation timing. Compare the printed drone with CAD and the training envelope before deploying a learned policy.

Written lesson + build labEvidence-based gate

The goal for this build

What you’ll be able to do.

  • Design measurements that can identify each claimed parameter.
  • Fit voltage-dependent thrust and actuator response.
  • Estimate inertia and account for fixture effects.
  • Propagate uncertainty into simulator and training coverage decisions.
What to have ready
  • Module 7 stable stock flight
  • Calibrated scale, secured thrust stand and suitable measurement fixtures
  • Module 5 randomization configuration
Download the course starter ↓

03 / Prove it

Keep the evidence.
Then run the gate.

What you ship

  • params_real.yaml with units, methods and uncertainty
  • Raw calibration, thrust, inertia and timing measurements
  • CAD-versus-real comparison and training-range coverage table
  • Measured-parameter benchmark and Module 08 evidence
Module 08 gateEvidence required
  • Every required parameter has a documented estimate and error bar or an explicit unresolved measurement limitation.
  • Measured intervals, and the assumed ranges of unresolved parameters, fit inside the training randomization ranges; mass and inertia are always measured.
  • The thrust map at the logged hover command matches the weighed m·g within 3%.
  • Any out-of-range value triggers a documented range update and retraining before Module 9.
  • Unresolved parameters are not silently converted into measured values.

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

The story behind the build

Episode companions.

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.