Video World Models · 2026-10-14

InfraWorld: Motion-Aware World Modeling for Roadside 4D Occupancy Forecasting

Songkai Wang, Xiaokai Bai, Lei Yang, Hui-liang Shen

arXiv:2610.14544PDF

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

TL;DR

InfraWorld is a roadside 4D occupancy forecasting world model that treats forecasting as selective state evolution: a dense BEV state keeps global scene context, while persistent motion queries model sparse localized dynamics. It uses Dynamic Query Initiation (DQI), Long–Short Temporal Fusion (LST), and Motion-Aware Query Injection (MAQI). Experiments on InfraOcc and Occ3D–nus report state-of-the-art performance; code will be released.

Why it matters

Roadside scenes are globally structured but evolve locally and sparsely. InfraWorld’s design assigns complementary roles to dense scene context and selective motion reasoning, aiming to preserve stable layout while focusing capacity on moving regions. The paper reports improved dynamic-category forecasting and maintains strong static-category performance, and also shows effectiveness under ego-centric settings.

Method

  • Formulate roadside 4D occupancy forecasting as selective state evolution with a dense BEV state for global scene context plus persistent motion queries for localized moving regions.
  • Use Dynamic Query Initiation (DQI) to initialize a limited set of persistent queries in motion-supported regions based on motion evidence from observed history.
  • During rollout, update the dense state with Long–Short Temporal Fusion (LST) and inject query-level dynamics into nearby BEV tokens using confidence-weighted Motion-Aware Query Injection (MAQI).

Limitation

The moving-mask head is supervised using a binary BEV target whose positive cells belong to dynamic semantic classes and have flow magnitude above a motion threshold; on InfraOcc, flow is generated offline from annotated 3D boxes and velocities and is used only after thresholding to identify moving occupancy cells.

Abstract (from arXiv)

Roadside 4D occupancy forecasting predicts future semantic occupancy in a fixed infrastructure coordinate system. Roadside scenes are globally structured yet locally evolving, requiring comprehensive scene-level context while precisely modeling sparse localized motion. However, existing forecasting paradigms struggle to tackle both requirements. To address this limitation, we propose InfraWorld, a world model formulating roadside occupancy forecasting as selective state evolution. InfraWorld maintains a dense BEV state for the complete scene and persistent motion queries for moving regions, assigning complementary states to global context and local dynamics. Specifically, we propose Dynamic Query Initiation (DQI) to initialize queries in regions supported by motion in the observed history. During rollout, the dense state is updated through Long--Short Temporal Fusion (LST), which combines short-term adjacent-frame continuity with long-term motion trends to preserve stable context while capturing scene evolution. Motion-Aware Query Injection (MAQI) then refines the queries and injects confidence-weighted query information into corresponding BEV tokens, enabling targeted dynamic updates without perturbing global context. Together, InfraWorld concentrates predictive capacity on evolving traffic regions. Extensive experiments on the InfraOcc and Occ3D--nus datasets demonstrate that InfraWorld achieves state-of-the-art performance for roadside 4D occupancy forecasting while retaining its effectiveness under ego-centric settings. Code will be released.

Related papers

Publication record

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

First public release: 2026-10-10 · Indexed by Pub-AI: 2026-10-14 UTC

preprint

InfraWorld: Motion-Aware World Modeling for Roadside 4D Occupancy Forecasting

Songkai Wang, Xiaokai Bai, Lei Yang, Hui-liang Shen

Preprint posted: 2026-10-10 · Original record ↗

arxiv
2610.14544
Explicit source versions (1)
  • v1 · 2026-10-10
  1. Preprint available

    Source date: 2026-10-10 · First detected: 2026-10-25 UTC · Historical baseline

    Source evidence (1)
Abstract · arxiv

Roadside 4D occupancy forecasting predicts future semantic occupancy in a fixed infrastructure coordinate system. Roadside scenes are globally structured yet locally evolving, requiring comprehensive scene-level context while precisely modeling sparse localized motion. However, existing forecasting paradigms struggle to tackle both requirements. To address this limitation, we propose InfraWorld, a world model formulating roadside occupancy forecasting as selective state evolution. InfraWorld maintains a dense BEV state for the complete scene and persistent motion queries for moving regions, assigning complementary states to global context and local dynamics. Specifically, we propose Dynamic Query Initiation (DQI) to initialize queries in regions supported by motion in the observed history. During rollout, the dense state is updated through Long--Short Temporal Fusion (LST), which combines short-term adjacent-frame continuity with long-term motion trends to preserve stable context while capturing scene evolution. Motion-Aware Query Injection (MAQI) then refines the queries and injects confidence-weighted query information into corresponding BEV tokens, enabling targeted dynamic updates without perturbing global context. Together, InfraWorld concentrates predictive capacity on evolving traffic regions. Extensive experiments on the InfraOcc and Occ3D--nus datasets demonstrate that InfraWorld achieves state-of-the-art performance for roadside 4D occupancy forecasting while retaining its effectiveness under ego-centric settings. Code will be released.