STAR-R1: Spacial TrAnsformation Reasoning by Reinforcing Multimodal LLMs

2025-05-21

AI Summary

To address the poor reasoning coherence and low exploration efficiency of multimodal large language models (MLLMs) on the cross-view object spatial transformation reasoning (TVR) task, this paper proposes the first single-stage reinforcement learning (RL) framework. Our method introduces a fine-grained reward function that explicitly models partial correctness, penalizes redundant enumeration, and suppresses passive non-responses—enabling, for the first time, human-like spatial comparison behavior. The framework jointly integrates MLLMs, TVR task formalization, and sparse-reward RL, eliminating the need for multi-stage training or costly human-annotated demonstration trajectories. Evaluated on 11 metrics, our approach achieves state-of-the-art (SOTA) performance across the board: cross-view accuracy improves by 23% over supervised fine-tuning, with significant gains in spatial relation identification and multi-object collaborative comparison capabilities.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Multi-modal Vision
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities across diverse tasks, yet they lag significantly behind humans in spatial reasoning. We investigate this gap through Transformation-Driven Visual Reasoning (TVR), a challenging task requiring identification of object transformations across images under varying viewpoints. While traditional Supervised Fine-Tuning (SFT) fails to generate coherent reasoning paths in cross-view settings, sparse-reward Reinforcement Learning (RL) suffers from inefficient exploration and slow convergence. To address these limitations, we propose STAR-R1, a novel framework that integrates a single-stage RL paradigm with a fine-grained reward mechanism tailored for TVR. Specifically, STAR-R1 rewards partial correctness while penalizing excessive enumeration and passive inaction, enabling efficient exploration and precise reasoning. Comprehensive evaluations demonstrate that STAR-R1 achieves state-of-the-art performance across all 11 metrics, outperforming SFT by 23% in cross-view scenarios. Further analysis reveals STAR-R1's anthropomorphic behavior and highlights its unique ability to compare all objects for improving spatial reasoning. Our work provides critical insights in advancing the research of MLLMs and reasoning models. The codes, model weights, and data will be publicly available at https://github.com/zongzhao23/STAR-R1.

Problem

• Addressing MLLMs' spatial reasoning gap vs humans
• Improving transformation identification in cross-view images
• Overcoming inefficient RL exploration in visual reasoning

Innovation

• Integrates single-stage RL with fine-grained rewards
• Rewards partial correctness, penalizes excessive enumeration
• Achieves state-of-the-art performance in spatial reasoning