Thinking with Imagination: Agentic Visual Spatial Reasoning with World Simulators

Hugging Face Daily Papers Papers
vision-language-models spatial-reasoning world-simulator reinforcement-learning agentic imagination

Summary

The paper proposes Astra, an agentic spatial reasoning framework that couples a reinforcement learning-trained VLM policy with a world simulator to generate novel-view observations for improved spatial reasoning in Vision-Language Models.

While Vision-Language Models (VLMs) have shown strong visual reasoning capabilities, their spatial reasoning abilities remain largely constrained to the observed images and text-oriented chain-of-thought. They often struggle to infer unobserved layouts, maintain cross-view consistency, and reason from alternative viewpoints when only limited egocentric observations are available. In this work, we study this problem as thinking with imagination, where a VLM actively acquires imagined visual evidence by interacting with a world simulator during reasoning. We propose Astra, an agentic spatial reasoning framework that empowers VLMs with action-conditioned visual imagination. Specifically, Astra couples Astra-VL, an RL-trained VLM policy, with Astra-WM, a Bagel-based world simulator that generates novel-view observations from context images and natural-language camera motions. To provide reliable imagined evidence, Astra-WM is trained with view consistency tuning to improve pose and content consistency across views. In the RL stage, we propose a world-simulator-in-the-loop two-phase RL curriculum to stabilize tool-use exploration and advance the model's ability to invoke the simulator only when imagined observations improve over direct answering. Experiments demonstrate that both the world simulator and the agentic policy are necessary: Astra-WM improves simulator-augmented Gemini-3-Flash on MMSI-Bench from 45.1 to 49.5, while Astra-VL improves the Qwen3-VL backbone from 29.8 to 38.8 on MMSI-Bench and from 36.8 to 42.7 on MindCube. These results show that imagined observations can provide useful spatial evidence, but effective world-model-augmented reasoning requires learning when, where, and how to imagine.
Original Article
View Cached Full Text

Cached at: 06/08/26, 07:14 AM

Paper page - Thinking with Imagination: Agentic Visual Spatial Reasoning with World Simulators

Source: https://huggingface.co/papers/2606.06476

Abstract

Astra is an agentic spatial reasoning framework that enhances Vision-Language Models with action-conditioned visual imagination by coupling a reinforcement learning-trained policy with a world simulator for generating novel-view observations.

WhileVision-Language Models(VLMs) have shown strong visual reasoning capabilities, theirspatial reasoningabilities remain largely constrained to the observed images and text-oriented chain-of-thought. They often struggle to infer unobserved layouts, maintain cross-view consistency, and reason from alternative viewpoints when only limited egocentric observations are available. In this work, we study this problem as thinking with imagination, where a VLM actively acquires imagined visual evidence by interacting with aworld simulatorduring reasoning. We propose Astra, an agenticspatial reasoningframework that empowers VLMs withaction-conditioned visual imagination. Specifically, Astra couples Astra-VL, an RL-trained VLM policy, with Astra-WM, aBagel-based world simulatorthat generates novel-view observations from context images and natural-language camera motions. To provide reliable imagined evidence, Astra-WM is trained withview consistency tuningto improve pose and content consistency across views. In the RL stage, we propose a world-simulator-in-the-loop two-phase RL curriculum to stabilizetool-use explorationand advance the model’s ability to invoke the simulator only when imagined observations improve over direct answering. Experiments demonstrate that both theworld simulatorand the agentic policy are necessary: Astra-WM improves simulator-augmented Gemini-3-Flash onMMSI-Benchfrom 45.1 to 49.5, while Astra-VL improves the Qwen3-VL backbone from 29.8 to 38.8 onMMSI-Benchand from 36.8 to 42.7 onMindCube. These results show that imagined observations can provide useful spatial evidence, but effective world-model-augmented reasoning requires learning when, where, and how to imagine.

View arXiv pageView PDFProject pageGitHub10Add to collection

Get this paper in your agent:

hf papers read 2606\.06476

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2606.06476 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2606.06476 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2606.06476 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

SpatialAct: Probing Spatial Reasoning-to-Action Capabilities of VLM Agents in 3D Scenes

Hugging Face Daily Papers

SpatialAct is a new simulator-grounded benchmark that probes whether VLM agents can perform coherent spatial reasoning and translate it into actions in 3D environments across multi-turn feedback settings. Experiments reveal a significant reasoning-to-action gap, with current VLMs struggling to maintain spatial beliefs and produce reliable actions despite performing well on isolated reasoning tasks.

SpatialASM: Internalizing Active Mentalization for Spatial Reasoning

Hugging Face Daily Papers

The paper introduces SpatialASM, a framework that trains large vision-language models to actively plan task-relevant viewpoints and mentally imagine missing spatial evidence via novel-view synthesis scaffolding, improving spatial reasoning across four backbones and seven benchmarks without extra views at inference.

Reinforcing Dual-Path Reasoning in Spatial Vision Language Models

Hugging Face Daily Papers

This paper introduces SR-REAL, a unified framework for spatial vision-language models that combines linguistic deduction and 3D geometric reasoning via reinforcement learning, enabling robust multi-step spatial reasoning across diverse tasks.