Research in end-to-end driving

Planning for the closed loop.

CLAP-AD explores imitation-free speed scoring to make end-to-end driving policies more aware of the outcomes their actions create.

WACV 2027 CARLA NAVSIM

Model releases

Checkpoints

Open weights, configs, and benchmark details.

CARLA · Bench2Drive-Speed

CLAP-AD TFv6

TFv6 checkpoint for closed-loop evaluation on Bench2Drive-Speed.

82.53 Speed-Adherence Score
View checkpoint and details →
NAVSIM v2.2 · navhard

CLAP-AD LTFv6

LTFv6 checkpoint adapted to NAVSIM's four-second planning horizon.

38.0 Overall EPDMS · two-stage protocol
View checkpoint and details →

Each model repository includes its checkpoint, configuration, evaluation details, intended-use notes, and license information. Scores follow the protocols documented in the model cards.

About the work

From imitation to outcome-aware planning.

CLAP-AD studies how speed selection and planning can account for closed-loop driving outcomes. The method and experiments span both CARLA and NAVSIM.

Intended use

Research artifacts

These checkpoints are for benchmark evaluation. They have not been validated for real-vehicle deployment or other safety-critical use.

Citation

Build on the research

@inproceedings{aghajanzadeh2027clapad,
  title={Closed-Loop Aware Planning via Imitation-Free Speed Scoring for End-to-End Autonomous Driving},
  author={Aghajanzadeh, Emad and Schwarzfischer, Thomas and Kaufeld, Marc and Thoma, Moritz and P{\'e}goud, Ryan and Vemparala, Manoj Rohit and Mori, Pierpaolo and Fasfous, Nael and Betz, Johannes and Mueller-Gritschneder, Daniel and Schlichtmann, Ulf},
  booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision},
  year={2027}
}