Disaster Rescue Rover System

A complete rescue rover stack that combines ESP32 firmware, a Raspberry Pi intelligence layer, a Go backend using Chi, and a React mission control dashboard. The rover explores, tracks its trail, detects victims, streams camera video, and reports real-time telemetry to a command center.

Architecture

• ESP32 firmware: motor control (L298N), IMU fusion (MPU6050), ultrasonic obstacle avoidance, GPS parsing (NEO-6M), and telemetry output over UART.
• Raspberry Pi: serial bridge, path tracking with GPS drift correction, person detection, and data relay to the backend.
• Go backend (Chi): REST API, WebSocket hub, and MJPEG camera stream proxy.
• React frontend: mission control UI with live map, telemetry, logs, alerts, and manual controls.

Data Flow

ESP32 -> Pi (serial) Pi -> Go (HTTP + WebSocket) Go -> React (WebSocket) React -> Go -> Pi -> ESP32 (commands)

Hardware (Strict)

• ESP32 DevKit V1
• L298N motor driver
• MPU6050 IMU
• HC-SR04 ultrasonic sensor
• NEO-6M GPS
• Raspberry Pi

Repository Layout

• backend/
  • main.go
  • router.go
  • websocket.go
  • telemetry_handler.go
  • models.go
  • camera.go
• frontend/
  • src/components/MapView.jsx
  • src/components/CameraFeed.jsx
  • src/components/TelemetryPanel.jsx
  • src/components/AlertsBar.jsx
  • src/components/LogsPanel.jsx
  • src/components/Controls.jsx
  • src/App.jsx
• pi/
  • serial_reader.py
  • path_tracker.py
  • drift_correction.py
  • vision.py
  • api_client.py
• esp32/
  • gps.cpp
  • imu.cpp
  • odometry.cpp
  • motor.cpp
  • sonar.cpp
  • telemetry.cpp
  • main.ino

Setup

1) ESP32 firmware

• Install Arduino IDE and ESP32 board support.
• Install library: TinyGPSPlus.
• Wiring defaults used in code:
  • Pi UART: RX=GPIO16, TX=GPIO17
  • GPS: RX=GPIO32, TX=GPIO4
  • Sonar: TRIG=GPIO5, ECHO=GPIO18
  • IMU: I2C SDA=GPIO21, SCL=GPIO22
  • L298N: IN1=GPIO27, IN2=GPIO26, IN3=GPIO25, IN4=GPIO33, ENA=GPIO14, ENB=GPIO12
• Flash esp32/main.ino

2) Raspberry Pi

sudo apt update
sudo apt install -y python3-pip python3-opencv
pip3 install pyserial requests websocket-client flask ultralytics

Run the bridge (includes MJPEG stream on port 8000):

python3 pi/serial_reader.py --port /dev/serial0 --backend http://<BACKEND_IP>:8080 --ws ws://<BACKEND_IP>:8080/ws --stream

3) Go backend

cd backend
go mod tidy
go run .

Environment options:

• PORT (default 8080)
• PI_CAMERA_STREAM_URL (default http://raspberrypi:8000/camera/stream)
• ALLOWED_ORIGINS (comma-separated, default *)

4) React dashboard

cd frontend
npm install
npm run dev

Optional environment overrides in frontend/.env:

VITE_BACKEND_HTTP_BASE=http://<BACKEND_IP>:8080
VITE_BACKEND_WS_URL=ws://<BACKEND_IP>:8080/ws

Notes

• GPS fixes are fused with odometry to reduce drift; without GPS, the map shows local coordinates only.
• Victim detection uses OpenCV HOG by default and can optionally switch to YOLO if a model is provided.
• The MJPEG stream is proxied through the Go backend at /camera/stream.