How joshuafernandes-code/guidebot works
GuideBot is a wearable navigation assistant: an ESP32 streams camera video to a laptop, where the main loop detects objects, estimates their distance and direction, and speaks hazard alerts. Voice commands can query nearby directions or request OCR and face identification. The README documents the streaming workflow and these major capabilities; wiring is based on the sampled main loop and modules, while peripheral hardware details are not shown.
Main entry points: main.py Runtime loop · stream.py Frame stream · depth.py Depth estimation · voice_input.py Voice commands
Components 14
Video capture
- ESP32 stream
esp32/guidebot_stream/guidebot_stream.ino - Frame stream
pipeline/stream.py
Laptop runtime
- Runtime loop
main.py - Configuration
config.py
Visual processing
- Object detection
pipeline/detection.py - Depth estimation
pipeline/depth.py - Hazard decisions
pipeline/decision.py - Text reading
pipeline/ocr.py - Face recognition
pipeline/face.py
Voice interaction
- Voice commands
audio/voice_input.py - Speech output
audio/tts.py
Other components
- Camera
- Wearer
- Audio output
Connections 19
- Camera → ESP32 stream: captures video
- ESP32 stream → Frame stream: streams video
- Runtime loop → Frame stream: requests frames
- Frame stream → Runtime loop: returns frames
- Runtime loop → Object detection: detects objects
- Object detection → Runtime loop: returns detections
- Runtime loop → Depth estimation: estimates depth
- Depth estimation → Runtime loop: returns depth map
- Runtime loop → Depth estimation: samples box distance
- Runtime loop → Hazard decisions: judges hazards
- Runtime loop → Hazard decisions: gets directions
- Runtime loop → Voice commands: polls commands
- Wearer → Voice commands: speaks
- Voice commands → Runtime loop: returns commands
- Runtime loop → Hazard decisions: answers queries
- Runtime loop → Text reading: reads frame
- Runtime loop → Face recognition: identifies faces
- Runtime loop → Speech output: queues speech
- Speech output → Audio output: speaks alerts