jbLambda is a ROS 2–based system for smartly triggering data recordings for autonomous driving. It lets teams deploy lightweight Python lambdas to vehicles that watch live sensor streams and trigger recordings when interesting events occur, reducing useless data capture and focusing storage on high‑value moments.
- Run edge lambdas that detect events (e.g., anomalies, near-misses, rare scenes)
- Trigger recordings with configurable pre/post‑roll windows
- Use ROS 2 topics and optional ONNX inference for decision logic
- Centrally deploy, update, and monitor lambdas across a fleet
jblambda-backend: API + WebSocket serverjblambda-frontend: management UIjblambda-launcher: edge agent that downloads and runs lambdasjblambda-rt: ROS 2 runtime that executes lambdasjblambda-lib: shared Rust types/utilitiesdocker/: container builds and deployment assets
See each package README for build/run details.
- Quickstart: build and run your first lambda
- Deployment Guide: deploy lambdas on your car via the framework
- Recipes: common trigger patterns and ONNX usage
- API Reference:
jbapifunctions and errors
The provided code and documentation was developed by the Deggendorf Insitute of Technology.
The development of this toolchain is funded by the German Federal Ministry for Economic Affairs and Energy within the project “just better DATA - Effiziente und hochgenaue Datenerzeugung für KI-Anwendungen im Bereich autonomes Fahren". The authors would like to thank the consortium for the successful cooperation.”
| JBDATA | BMWE |
|---|---|
 |
 |