Projects
All projects follow the Physical AI system architecture:
Sensor → Model → Decision → Actuator
Day 1 ends with a guided project (done together as a group). Day 2 is your team's own build — pick one of the ideas below or propose your own.
Guided Project (Day 1)
The facilitator leads the full group through one complete Physical AI project from start to finish — data collection, model training on Edge Impulse, deployment to Arduino Q, and actuator response. This is your reference for Day 2.
Project TBD based on batch — confirmed on Day 1.
Project Ideas for Day 2
🎤 Gesture / Motion Classifier
Sensors: IMU (Accelerometer + Gyro)
Model: Motion classification — TinyML on Edge Impulse
Actuator: LED, Servo, Buzzer
Idea: Classify specific hand gestures or arm movements and trigger a corresponding response (rotate servo, change LED colour, sound alert).
👁️ Visual Classifier
Sensors: USB Camera
Model: Image classification — Edge Impulse FOMO or MobileNet
Actuator: LED, Relay, Buzzer
Idea: Detect a specific object, person presence, or condition from camera feed and trigger an action.
🔊 Sound Classifier
Sensors: USB Camera with built-in mic, or Microphone Module (if camera unavailable)
Model: Sound classification — keyword spotting or event detection
Actuator: LED, Servo, Buzzer
Idea: Detect a specific sound (clap, voice command, alarm) and respond with an actuator.
🤖 Intelligent Robot (Obstacle Avoidance + AI)
Sensors: Ultrasonic Sensor (HC-SR04), IMU
Model: Obstacle detection / behaviour classification
Actuator: 2WD Robot Chassis, DC Motors, Motor Driver
Idea: A robot that navigates a path using distance sensing and an AI decision layer.
🚗 Driver Drowsiness Monitor (Community Suggestion)
Sensors: USB Camera
Model: Eye-state or head-pose classification
Actuator: Buzzer alert
Idea: Detect when a driver's eyes are closed or head is drooping and trigger an alert. (Autonomous Car Monitoring — driver sleeping use case.)
📋 AI-Based Attendance System (Community Suggestion)
Sensors: USB Camera
Model: Face detection / presence classification
Actuator: OLED display, LED indicator
Idea: Detect presence of a person and log attendance automatically using an on-device vision model.
🌡️ Anomaly Detection on Sensor Data
Sensors: Temperature Sensor, Vibration Sensor
Model: Anomaly detection autoencoder — Edge Impulse
Actuator: Buzzer, LED alert
Idea: Learn what "normal" looks like and flag deviations — mimics industrial predictive maintenance.
Choosing Your Project
| Project | Sensors Needed | Difficulty |
|---|---|---|
| Gesture Classifier | IMU (built-in on Arduino Q) | ★☆☆ Beginner |
| Sound Classifier | USB Camera / Mic Module | ★☆☆ Beginner |
| Visual Classifier | USB Camera | ★★☆ Intermediate |
| Anomaly Detection | Temp + Vibration Sensors | ★★☆ Intermediate |
| Intelligent Robot | Ultrasonic + Motor Kit | ★★★ Advanced |
| Driver Drowsiness | USB Camera | ★★☆ Intermediate |
| Attendance System | USB Camera | ★★☆ Intermediate |
:::tip Recommended for First-Timers Start with the Gesture Classifier — it uses only the built-in IMU on the Arduino Q (no extra wiring), and the full pipeline (data → train → deploy → demo) takes about 90 minutes. :::
Final Deliverables Checklist
Each team submits:
- Project documentation (hackster.io or Arduino Project Hub)
- Source code (GitHub link)
- Dataset summary (Edge Impulse project link or export)
- Demo video or photos
- Brief presentation (live demo on Day 2)
Next Steps
→ Bootcamp Inventory — check what hardware is available for your chosen project
→ Arduino Q First Project — get comfortable with the board before Day 2 build