# GuidedWAM: Training-Free Energy Guidance for World Action Models in Autonomous Driving

DevFeed: [GuidedWAM: Training-Free Energy Guidance for World Action Models in Autonomous Driving](<https://devfeed.tech/articles/guidedwam-training-free-energy-guidance-for-world-action-models-in-autonomous-driving-120266.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.07912>)

Author: Xiyan Su, Finn Rasmus Sch\\"afer, Hojun Lim, Frank Diermeyer, Johannes Betz, Markus Lienkamp

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [World models](<https://devfeed.tech/topics/world-models.md>), [autonomous vehicles](<https://devfeed.tech/topics/autonomous-vehicles.md>)

Tags: [autonomous-driving](<https://devfeed.tech/tags/autonomous-driving.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [energy](<https://devfeed.tech/tags/energy.md>), [guidance](<https://devfeed.tech/tags/guidance.md>), [models](<https://devfeed.tech/tags/models.md>), [training](<https://devfeed.tech/tags/training.md>), [world-model](<https://devfeed.tech/tags/world-model.md>)

## AI overview

GuidedWAM adds driving objectives to frozen world action models at sampling time, avoiding retraining. It represents objectives as differentiable energies on the sampler's intermediate clean-action estimate and requires only forward access to the frozen model. Sampling-time scheduling, per-energy normalization, and velocity-bounded correction stabilize the guidance. On the NAVSIM v2 hard split, it improved all three tested publicly released models by 1.8-3.6 EPDMS, averaged across three sampler seeds, with less than 10% added sampling time per step.

## Source excerpt

arXiv:2611.07912v1 Announce Type: new 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 into 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 3.6 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.