A spatial memory assistant that helps you find objects in your environment using computer vision and 3D mapping.

• Map your space — Walk around with a camera to build a 3D map of your environment
• Detect objects — Automatically recognizes and remembers object locations using YOLOv8
• Find things — Ask "Where are my keys?" and get voice-guided directions to the object

Frontend: Next.js, React Three Fiber, Tailwind CSS
Backend: ROS2, OpenCV, ONNX Runtime (YOLOv8)
APIs: Cerebras (LLM), ElevenLabs (TTS)

cd frontend/app
npm install
npm run dev

cd ros2
colcon build
source install/setup.bash
ros2 launch cloud_web_bridge depthai.launch.py

Create frontend/app/.env.local:

CEREBRAS_API_KEY=your_key
ELEVENLABS_API_KEY=your_key
NEXT_PUBLIC_ROSBRIDGE_URL=ws://localhost:9090

├── frontend/app/     # Next.js web app
└── ros2/src/         # ROS2 packages
    ├── cloud_web_bridge/    # WebSocket bridge
    ├── depth_mapping/       # Spatial recognition
    └── luxonis_camera/      # Camera driver

MIT