Reference Autonomy Stack (Engineering Harness)

We keep the focus on engineering: deterministic behavior, measurable safety metrics, and reproducible validation. This reference autonomy service exists to make end-to-end demos and integration tests easy — not to replace an OEM stack.

Local ports

  • 8000: Perception / FieldSpace API (`api_server.py`)
  • 8001: Autonomy API (`l5_api_server.py`)
  • 8888: Reference frontend in the ADS repo (this website may run on 3000)

Key endpoints

POST /api/l5/initialize POST /api/l5/{session_id}/engage POST /api/l5/{session_id}/disengage POST /api/l5/{session_id}/emergency-stop POST /api/l5/{session_id}/process GET /api/l5/{session_id}/status GET /api/l5/{session_id}/trajectory GET /api/l5/{session_id}/metrics WS /ws/l5/{session_id}

On this website, HTTP calls can go through /api/l5proxy. WebSockets connect directly to the autonomy server.

Use it from the Developers dashboard

  1. Start the autonomy backend: POST /api/devtools/start-l5 (button in the dashboard)
  2. Open /developers?tab=l5 to initialize a session and connect WebSocket updates.