# Trajectory Evaluation via Rollout-Free World Model for End-to-end Autonomous Driving

DevFeed: [Trajectory Evaluation via Rollout-Free World Model for End-to-end Autonomous Driving](<https://devfeed.tech/articles/trajectory-evaluation-via-rollout-free-world-model-for-end-to-end-autonomous-driving-117559.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.05072>)

Author: Junzhe Wang, Jinqing Zhang, Qingjie Liu

Published: 2026-11-04T05:00:00Z

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](<https://devfeed.tech/tags/autonomous.md>), [autonomous-driving](<https://devfeed.tech/tags/autonomous-driving.md>), [candidate](<https://devfeed.tech/tags/candidate.md>), [comparison](<https://devfeed.tech/tags/comparison.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [driving](<https://devfeed.tech/tags/driving.md>), [end-to-end-autonomous-driving](<https://devfeed.tech/tags/end-to-end-autonomous-driving.md>), [errors](<https://devfeed.tech/tags/errors.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [future](<https://devfeed.tech/tags/future.md>), [horizon](<https://devfeed.tech/tags/horizon.md>), [inference](<https://devfeed.tech/tags/inference.md>), [information](<https://devfeed.tech/tags/information.md>), [lightweight](<https://devfeed.tech/tags/lightweight.md>), [memory](<https://devfeed.tech/tags/memory.md>), [model](<https://devfeed.tech/tags/model.md>), [overhead](<https://devfeed.tech/tags/overhead.md>), [world](<https://devfeed.tech/tags/world.md>), [world-model](<https://devfeed.tech/tags/world-model.md>)

## AI overview

RFWorld evaluates candidate driving trajectories without predicting a separate future-state rollout for each one. It builds a shared scene memory, shaped during training by lightweight temporal readout with semantic supervision; at inference, each candidate sparsely queries that memory for scoring. The authors say this reduces computation by roughly the planning horizon multiplied by the number of candidates and avoids recursive rollout errors. Experiments on NAVSIM and Bench2Drive are reported to show state-of-the-art performance and a more planning-aligned representation, but the excerpt gives no metrics or comparison details.

## Source excerpt

arXiv:2611.05072v1 Announce Type: cross Abstract: Although existing candidate trajectory evaluators have substantially improved end-to-end planning accuracy, their underlying mechanism remains suboptimal. Model-based evaluators typically make scoring future-aware by explicitly predicting candidate-conditioned future states over the planning horizon, which we term rollout. However, rollout incurs high inference cost and accumulates prediction errors. We argue that planning does not require reconstructing a full future scene for each candidate, but only future-relevant information for reliable comparison. We therefore propose RFWorld, a Rollout-Free World model for end-to-end trajectory evaluation. RFWorld constructs a shared scene memory and shapes it into a future-readable representation via lightweight training-time temporal readout with semantic supervision, keeping training overhead low. At inference time, candidate trajectories directly query the shared memory via sparse spatial readout for scoring, reducing computation by roughly the planning horizon times the number of candidate trajectories, while avoiding recursive rollout errors. Experiments on NAVSIM and Bench2Drive show that RFWorld learns a more planning-aligned world representation and achieves state-of-the-art performance.