WebRTC Car Video Prototype

Repository contains 3 services:

• signaling-server (Node.js): auth + signaling relay
• frontend (Vite/TS): operator UI
• car-video-client (Python): car-side WebRTC sender

Unified start

python start.py

Optional:

python start.py --skip-update

Wrappers:

• Linux/macOS: ./start.sh
• Windows: start.bat

start.py creates missing .env files from .env.example, installs dependencies (unless skipped), picks free ports, and runs all 3 services.

Video source modes

car-video-client supports only:

• RTSP cameras (MEDIA_SOURCES list, multiple per car)
• local webcams (MEDIA_SOURCES with type=webcam)

After selecting a car, frontend shows all of its camera tracks (up to 4 tiles now). On source failures it uses a placeholder image (car-video-client/assets/no_signal.png).

Security baseline

• Set signaling-server/.env: ALLOW_INSECURE_TOKENS=false
• Use reachable Keycloak host/IP consistently in all services (for local: http://127.0.0.1:8080)
• Configure:
  • KEYCLOAK_JWKS_URL / KEYCLOAK_ISSUER in signaling
  • KEYCLOAK_TOKEN_URL + client credentials in car client
  • VITE_KEYCLOAK_URL in frontend

Quick demo mode (no Keycloak)

• signaling-server/.env: ALLOW_INSECURE_TOKENS=true
• frontend/.env: VITE_SKIP_KEYCLOAK=true
• car-video-client/.env: KEYCLOAK_TOKEN_URL= and SIGNALING_AUTH_TOKEN=demo