# 📷 Lesson Plan — Vision Lab

> **Ages 11-14 · 50 min · 1 robot per pair · webcam-equipped device**

## 🎯 Objective
Students drive the robot using their face or pose tracked by the device camera, see what an **AI vision model** can and cannot do, and discuss privacy + bias.

## 🧰 Materials
- 1 Maqueen Lite v4 robot per pair (charged)
- 1 laptop or tablet **with built-in camera** per pair (Chrome or Edge)
- Stable lighting (avoid backlight)
- Optional: simple props (hat, scarf) to test the model's robustness

## 📋 Lesson flow (50 min)

| Time | Phase | Teacher | Students |
|---|---|---|---|
| 0-5 | **Hook** | Show a smartphone face-unlock fail. "When does the AI mess up?" | Discuss: glasses, low light, twins. |
| 5-15 | **Demo** | Open Vision Lab. Connect camera. Tilt head left → robot turns left. Show **face mode** then **pose mode**. | Watch; predict what kind of head movement maps to what motion. |
| 15-30 | **Exploration** | Pairs connect, calibrate face neutral position, try a 1-minute drive without hands. | Drive the robot; trade who pilots. |
| 30-40 | **Challenge** | "Drive a figure-8 path using only your face." | Plan moves; document misses. |
| 40-48 | **Discussion** | Open data/privacy thread: where is the camera image processed? (Answer: 100% in the browser, never sent.) "What if it wasn't?" | Debate; share examples of biased/unfair AI from real life. |
| 48-50 | **Reflection** | Whip-around: "Is AI smart, or fast and lucky?" | Articulate one nuance about how the model "sees". |

## ✅ Success criteria
- I can calibrate my face's neutral position.
- I can drive a controlled path using only my face.
- I can explain in one sentence where the camera image goes.

## 📚 Standards
- **FR Cycle 3** — *EMI* : usages responsables du numérique ; biais des algorithmes.
- **FR Cycle 4** — *Tech* : intelligence artificielle (introduction). *EMI* : protection vie privée, traitement d'image.
- **NGSS** — MS-ETS1-3 (analyze data from tests).
- **CSTA** — 2-AP-15 (apply software dev best practices); 2-IC-20 (computing impacts).

## 🚀 Extension
- Switch to pose mode and have the robot mirror full-body gestures.
- Compare with **Co-Pilot Lab**: voice vs. vision — which is more reliable, when?
- Research: where is face-recognition used / abused in your country?