Driving · 2026-10-14
OccWAM: Occupancy World-Action Models with Joint Scene-to-Ego Flow Matching
Qihang Huang, Wenhui Huang, Collister Chua, Shiyu Fang, Zhan Chen, Songyan Zhang, Zhidong Wang, Daiying Tian, Chen Lv
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
OccWAM is an occupancy-native world–action model that jointly generates dense metric future occupancy and continuous ego trajectories using scene-to-ego two-stage conditional flow matching. Stage I predicts scene-centric occupancy plus an absolute ego trajectory in a fixed current-ego-anchored frame; Stage II re-indexes the predicted world into future ego frames and jointly refines ego-centric occupancy and motion for action-aligned planning. It avoids pixel synthesis and preserves semantic geometry.
Why it matters
The authors argue that existing occupancy-aware planners impose an explicit coupling order or role asymmetry between world prediction and action generation, limiting reciprocal interaction. OccWAM’s coupled conditional flows let occupancy directly participate in generative planning, producing an independently measurable, geometrically aligned future world (occupancy) without pixel-space synthesis, aimed at driving and planning tasks on Bench2Drive and nuScenes.
Method
- Occupancy-native world–action generation: jointly transports dense future occupancy latents and continuous ego trajectory through coupled conditional flows conditioned on scene representations, using bidirectional interaction between World and Action experts at every transformer layer.
- Scene-to-ego two-stage flow matching: Stage I generates scene-centric occupancy and an absolute ego trajectory in a fixed current-ego-anchored frame; the predicted world is then re-indexed into future ego frames induced by the Stage-I trajectory and converted to adjacent pose increments for Stage II.
- Semantic geometry without pixel synthesis: uses dense metric occupancy (via a frozen frame-wise occupancy autoencoder) for direct planning reasoning and occupancy supervision, rather than video or pixel-space futures.
Abstract (from arXiv)
Occupancy world models provide a structured, metric representation of future driving scenes, making them a promising basis for end-to-end autonomous driving. However, existing occupancy-aware planners typically couple occupancy prediction and action generation through either hierarchical coupling or parallel multi-task learning. Hierarchical methods use occupancy either as a fully predicted future preceding action generation or as an action-conditioned rollout for trajectory evaluation, whereas parallel multi-task methods predict occupancy alongside planning through a separate branch or auxiliary objective. These interfaces retain either an explicit generation order or a role asymmetry, limiting reciprocal interaction between world prediction and action generation during trajectory formation. We propose OccWAM, an occupancy-native world-action model that jointly generates dense future occupancy and continuous ego trajectories via flow matching conditioned on scene representations. Building on this joint formulation, we introduce scene-to-ego world-action flow matching, a two-stage formulation. The first stage predicts scene-centric occupancy together with an absolute ego trajectory, providing a stable representation of future scene evolution. The same predicted world is then re-indexed into the future ego frames induced by the initial trajectory, where ego motion is represented through adjacent pose increments and a second joint flow refines ego-centric occupancy and motion for action-aligned planning. Unlike video-based world-action models, OccWAM preserves explicit semantic geometry for direct planning reasoning without pixel synthesis. Extensive experiments on Bench2Drive and nuScenes demonstrate state-of-the-art performance in closed-loop driving, open-loop planning, and future occupancy prediction.
Related papers
- PhysWAM: Physically Consistent World Action Model for Autonomous Driving
- AffordDrive3D: Affordance-Aware World-Action Modeling with Spatial Understanding
- PlanWAM: Planning-Shaped Future Representations for End-to-End Autonomous Driving
- VGGTWorld-VLA: Intent-Conditioned 3D World Evolution for Autonomous Driving
- Explicit Geometric Chain-of-Thought for Vision-Language-Action in Autonomous Driving
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-13 · Indexed by Pub-AI: 2026-10-14 UTC
OccWAM: Occupancy World-Action Models with Joint Scene-to-Ego Flow Matching
Preprint posted: 2026-10-13 · Original record ↗
- arxiv
- 2610.17688
Explicit source versions (1)
- v1 · 2026-10-13
- Preprint available
Source date: 2026-10-13 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
Occupancy world models provide a structured, metric representation of future driving scenes, making them a promising basis for end-to-end autonomous driving. However, existing occupancy-aware planners typically couple occupancy prediction and action generation through either hierarchical coupling or parallel multi-task learning. Hierarchical methods use occupancy either as a fully predicted future preceding action generation or as an action-conditioned rollout for trajectory evaluation, whereas parallel multi-task methods predict occupancy alongside planning through a separate branch or auxiliary objective. These interfaces retain either an explicit generation order or a role asymmetry, limiting reciprocal interaction between world prediction and action generation during trajectory formation. We propose OccWAM, an occupancy-native world-action model that jointly generates dense future occupancy and continuous ego trajectories via flow matching conditioned on scene representations. Building on this joint formulation, we introduce scene-to-ego world-action flow matching, a two-stage formulation. The first stage predicts scene-centric occupancy together with an absolute ego trajectory, providing a stable representation of future scene evolution. The same predicted world is then re-indexed into the future ego frames induced by the initial trajectory, where ego motion is represented through adjacent pose increments and a second joint flow refines ego-centric occupancy and motion for action-aligned planning. Unlike video-based world-action models, OccWAM preserves explicit semantic geometry for direct planning reasoning without pixel synthesis. Extensive experiments on Bench2Drive and nuScenes demonstrate state-of-the-art performance in closed-loop driving, open-loop planning, and future occupancy prediction.