Driving · 2026-10-28

Closed-Loop Aware Planning via Imitation-Free Speed Scoring for End-to-End Autonomous Driving

Emad Aghajanzadeh, Thomas Schwarzfischer, Marc Kaufeld, Moritz Thoma, Ryan P\'egoud, Manoj Rohit Vemparala, Pierpaolo Mori, Nael Fasfous, Johannes Betz, Daniel Mueller-Gritschneder, Ulf Schlichtmann

arXiv:2610.36079 PDF

AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.

TL;DR

CLAP-AD is a plug-in planning head for end-to-end autonomous driving that learns path by imitation but learns speed by imitation-free multi-criteria scoring (collision, traffic light, stop sign) under physically grounded safety constraints. It enables explicit test-time control of speed selection without speed-conditioned training by scoring candidate speed classes and selecting a feasible one.

Why it matters

The authors argue that imitation of a single expert speed is inadequate for driving’s multi-modal nature and prevents adapting to user-specified speed preferences at test time. By separating path imitation from speed scoring and adding deployment-time speed selection, CLAP-AD targets controllability while keeping safety criteria explicit and interpretable.

Method

CLAP-AD integrates with existing end-to-end driving models: an Ego-Path block predicts a driving path, then a speed scoring head evaluates candidate speed classes along the predicted path using separate branches for collision, traffic lights, and stop signs.
Speed scoring is trained with a multi-criteria, imitation-free objective derived from logged open-loop data and physically grounded safety rules, rather than direct imitation of expert speed.
At deployment, candidate speeds are deemed feasible via a threshold over safety scores; the target speed is selected among feasible classes (e.g., closest to the user request), and the system falls back to 0 m/s if no class is feasible.

Limitation

The scoring in CLAP-AD covers three criteria—collision, traffic lights and stop signs—and does not consider road speed limits (e.g., on highways). The candidate speed set is chosen to cover typical urban driving, and higher-speed settings would require an extended speed range and a criterion for applicable speed limits.

Abstract (from arXiv)

End-to-end autonomous driving has predominantly relied on open-loop imitation learning for training, which presents two limitations. First, expert demonstrations show only one behavior per scenario, so the model cannot learn alternative valid behaviors and, in turn, cannot adapt to user-specified preferences at test time. Second, noise or ambiguity in these demonstrations can produce inconsistent, unreliable supervision for the end-to-end models. Driving is inherently multi-modal: multiple different behaviors can be valid responses to the same scene, and this diversity arises largely from speed rather than path. We introduce a planning paradigm that learns the path through imitation while scoring a set of candidate speeds across multiple criteria with a learned function supervised by physically grounded safety constraints. We realize this paradigm as Closed-Loop Aware Planning for Autonomous Driving (CLAP-AD), a plug-in planning head that integrates with existing end-to-end driving models and, to our knowledge, is the first to enable explicit test-time control over their speed selection without speed-conditioned training. On Bench2Drive-Speed, CLAP-AD sets a new state of the art Speed-Adherence Score of 82.53, topping TFv6, the strongest baseline on this metric, by 12.9 points while retaining strong closed-loop driving performance. On Bench2Drive, it improves driving score by +28.5 and +32.7 over UniAD and VAD, and on NAVSIM v2, built on real-world driving data, EPDMS by +6.1 over its backbone. Project page: https://clap-ad.github.io

Publication record

Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.

First public release: 2026-10-26 · Indexed by Pub-AI: 2026-10-28 UTC

preprint