Part 5 · Perception and estimation
Turning noisy sensors and camera pixels into a state the controller can trust.
Learn first
Section titled “Learn first”This part assumes the following. If any of it is new, learn it on AI Engineering from Scratch first:
- Probability and Distributions
- Bayes’ Theorem
- Stochastic Processes
- Image Fundamentals — Pixels, Channels, Color Spaces
- Object Detection — YOLO from Scratch
- Open-Vocabulary Vision — CLIP
- SAM 3 & Open-Vocabulary Segmentation
- Monocular Depth & Geometry Estimation
Lessons
Section titled “Lessons”| # | Lesson | Run type | Time | Status |
|---|---|---|---|---|
| 17 | Sensors, noise and filtering | CPU | ~45 min | ✅ done |
| 18 | The Kalman filter | CPU | ~60 min | ✅ done |
| 19 | Camera model, calibration and hand-eye calibration | CPU | ~60 min | ✅ done |
| 20 | From pixels to the table: homography and depth back-projection | CPU | ~45 min | ✅ done |
| 21 | Object perception for manipulation | CPU | ~60 min | ✅ done |