Driving · 2026-10-14
G$^2$WAM: World Action Models for Efficient and Geometry Grounded Image-Goal Navigation
Ke Gui, Chenghao Shi, Ziyue Wang, Longwen Chen, Xieyuanli Chen, Huimin Lu
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
G$^2$WAM (Geometry Grounded World Action Model) targets image-goal navigation where the goal is often out of view. It anchors current, goal, and history frames in a shared scene frame using a learned scene-coordinate localization module, providing stable relative pose. Using the world-action-model paradigm, it jointly predicts latent future states and canonical waypoint increments entirely in latent space, avoiding CEM-style planning and pixel reconstruction.
Why it matters
The authors address two bottlenecks in image-goal navigation: (1) keeping consistent spatial context over a navigation history without expensive 3D reconstruction, and (2) reducing inference cost by removing CEM-style sampling/selection. Their geometry-grounded relative-pose conditioning and latent-space world-action rollouts aim to improve long-horizon trajectory prediction and closed-loop navigation efficiency.
Method
- Anchors observation history and goal image in a shared scene frame via an end-to-end localization module that estimates current/goal/history poses, yielding the current-to-goal relative transformation and history motion increments.
- Uses a world-action-model design that jointly predicts latent future states and canonical waypoint increments, eliminating CEM-style planning; rollouts are performed entirely in latent space without pixel-level reconstruction.
- Conditions the latent transition on relative pose (goal geometry) and a motion profile (history), and adapts canonical-motion predictions to executable platform actions via a canonical-motion adapter.
Limitation
Scene-coordinate anchoring assumes a previously seen scene, as it requires pre-collected trajectory data.
Abstract (from arXiv)
World models capture environment dynamics and support counterfactual reasoning for planning. In image-goal navigation, where the target often lies beyond the current field of view, the agent must not only model environmental dynamics but also maintain a persistent spatial memory of the scene. Existing approaches either introduce spatial context by fine-tuning pretrained models on the target task, which underexploits their spatial priors, or rely on external 3D reconstruction, which is computationally expensive for real-world deployment. In this paper, we propose the Geometry Grounded World Action Model (G$^2$WAM) for image-goal navigation. G$^2$WAM encodes spatial cues through a learned scene-coordinate localization module that anchors the current, goal, and history frames in a common reference frame, yielding a stable and consistent spatial context that enables efficient navigation without explicit 3D reconstruction. Following the world-action-model paradigm, G$^2$WAM jointly predicts future states and actions, eliminating the need for costly cross-entropy method (CEM) style planning. In addition, G$^2$WAM operates entirely in latent space, avoiding the computational burden of pixel-level reconstruction and substantially reducing inference cost. Across four trajectory prediction benchmarks, G$^2$WAM reduces average ATE and RPE by 40.1% and 35.2%, respectively. In closed-loop navigation experiments, it improves the success rate from 54.0% to 63.0% and the success weighted by path length (SPL) from 36.4% to 59.2%, while maintaining high efficiency.
Related papers
- SkipWorld: Latent Skip Planning with Distilled World Models for Efficient Visual Navigation
- {\Delta}WAM: Distilling Action Tangent Fields into World Action Models
- LiteNWM: Efficient Latent World Models for Onboard Visual Navigation in the Wild
- DiffWAM: A Fast and Efficient Navigation World Action Model
- GeoWM: Efficient Direct World Modeling in Explicit Geometry
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-12 · Indexed by Pub-AI: 2026-10-14 UTC
G$^2$WAM: World Action Models for Efficient and Geometry Grounded Image-Goal Navigation
Preprint posted: 2026-10-12 · Original record ↗
- arxiv
- 2610.16558
Explicit source versions (1)
- v1 · 2026-10-12
- Preprint available
Source date: 2026-10-12 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
World models capture environment dynamics and support counterfactual reasoning for planning. In image-goal navigation, where the target often lies beyond the current field of view, the agent must not only model environmental dynamics but also maintain a persistent spatial memory of the scene. Existing approaches either introduce spatial context by fine-tuning pretrained models on the target task, which underexploits their spatial priors, or rely on external 3D reconstruction, which is computationally expensive for real-world deployment. In this paper, we propose the Geometry Grounded World Action Model (G$^2$WAM) for image-goal navigation. G$^2$WAM encodes spatial cues through a learned scene-coordinate localization module that anchors the current, goal, and history frames in a common reference frame, yielding a stable and consistent spatial context that enables efficient navigation without explicit 3D reconstruction. Following the world-action-model paradigm, G$^2$WAM jointly predicts future states and actions, eliminating the need for costly cross-entropy method (CEM) style planning. In addition, G$^2$WAM operates entirely in latent space, avoiding the computational burden of pixel-level reconstruction and substantially reducing inference cost. Across four trajectory prediction benchmarks, G$^2$WAM reduces average ATE and RPE by 40.1% and 35.2%, respectively. In closed-loop navigation experiments, it improves the success rate from 54.0% to 63.0% and the success weighted by path length (SPL) from 36.4% to 59.2%, while maintaining high efficiency.