GuidedWAM: Training-Free Energy Guidance for World Action Models in Autonomous Driving
GuidedWAM: Training-Free Energy Guidance for World Action Models in Autonomous Driving
This paper introduces GuidedWAM, a model-agnostic framework that enables injecting driving objectives into frozen World Action Models (WAMs) without retraining. By constructing objectives as differentiable energies on intermediate action estimates and applying sampling-time guidance scheduling, normalization, and velocity-bounded correction, the method improves performance across three WAMs on the NAVSIM v2 hard-split by 1.8 to 3.6 EPDMS while adding less than 10% to sampling time per step.
The authors propose GuidedWAM, a training-free framework that injects driving objectives into frozen World Action Models (WAMs) via energy guidance at sampling time. On the NAVSIM v2 hard-split benchmark, this approach improves three publicly released WAMs by 1.8 to 3.6 EPDMS on average while increasing sampling time by less than 10% per step.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
to frozen WAMs at sampling time. Our framework constructs each objective as differentiable energies on the sampler's intermediate clean-action estimate, which needs only forward access to the frozen model, and is stabilized by sampling-time guidance scheduling, per-energy normalization, and velocity-bounded correction. On the NAVSIM v2 hard-split, guidance improves all three frozen, publicly released WAMs by 1.8 to 3
Open source S3
.6 EPDMS, averaged over three sampler seeds, and adds less than 10% to the sampling time per step. These results position energy guidance as a practical post-training control channel for WAMs.
Open source S4
arXiv:2611.07912v1 Announce Type: cross Abstract: World action models (WAMs) for autonomous driving couple the driving policy and the visual world model in a single generative network whose objectives are fixed post-training. Conditioning on a new driving objective requires retraining. We study energy guidance as a general mechanism and present GuidedWAM, a model-agnostic framework for injecting driving objectives in
Open source S2
Source：arXiv Robotics — research abstracts · arxiv.org