SpatialReasoner: Towards Explicit and Generalizable 3D Spatial Reasoning

2025-04-28

AI Summary

Existing 3D spatial reasoning methods predominantly rely on implicit representations, lacking systematic evaluation of generalization capability across diverse 3D knowledge domains. Method: We propose the first vision-language large model (LVLM) explicitly designed for interpretable and generalizable 3D spatial reasoning. Its core innovation is the introduction of a unified, explicit 3D symbolic representation as a shared interface across perception, computation, and reasoning stages. This enables error attribution analysis and reasoning-path diagnosis. The architecture integrates a 3D perception encoder, a geometry-aware spatial computation module, and an RL-optimized symbolic reasoning framework, enabling end-to-end interpretable 3D understanding. Contribution/Results: Our model achieves state-of-the-art performance across multiple benchmarks, improves zero-shot generalization accuracy on unseen 3D question types by 12.6%, and enables precise localization of spatial factual errors.

Technology Category

• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Computer Vision: Visual Reasoning & Symbolic Representations
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Large language models for search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Recent studies in 3D spatial reasoning explore data-driven approaches and achieve enhanced spatial reasoning performance with reinforcement learning (RL). However, these methods typically perform spatial reasoning in an implicit manner, and it remains underexplored whether the acquired 3D knowledge generalizes to unseen question types at any stage of the training. In this work we introduce SpatialReasoner, a novel large vision-language model (LVLM) that address 3D spatial reasoning with explicit 3D representations shared between stages -- 3D perception, computation, and reasoning. Explicit 3D representations provide a coherent interface that supports advanced 3D spatial reasoning and enable us to study the factual errors made by LVLMs. Results show that our SpatialReasoner achieve improved performance on a variety of spatial reasoning benchmarks and generalizes better when evaluating on novel 3D spatial reasoning questions. Our study bridges the 3D parsing capabilities of prior visual foundation models with the powerful reasoning abilities of large language models, opening new directions for 3D spatial reasoning.

Problem

• Addresses implicit 3D spatial reasoning in current methods
• Explores generalizability of 3D knowledge to unseen questions
• Bridges 3D parsing and reasoning in vision-language models

Innovation

• Uses explicit 3D representations for reasoning
• Integrates perception, computation, and reasoning stages
• Combines vision-language model with 3D parsing