🚗 A fast and flexible planning framework combining configuration and trajectory streams for real-time maneuver planning.
- Hybrid planning with fast heuristics
- Trajectory prediction using QCNet
- Integration with CommonRoad benchmark
- Apollo-compatible trajectory publishing
ffstreams/: core FFStreams++ implementation (Python)ffplanner/: FastForward heuristic planner on PDDL2.1 (C++)auto_driving/: different maneuvers PDDL2.1 domainsscenarios/: benchmark scripts for CommonRoadmetrics/: metrics for evaluation (OPM,TTC)prediction/: QCNet Python inference
- Weighted A* on configuration graph
- Trajectory stream refinement in Frenet
- Prediction-aware planning with uncertainty
Jamal, M. et al. FFStreams: Fast Search With Streams for Autonomous Maneuver Planning. IEEE RA-L, 2024.
