AIMar 25, 2025

LEGO-Puzzles: How Good Are MLLMs at Multi-Step Spatial Reasoning?

Peking U
arXiv:2503.19990v344 citationsh-index: 33
Originality Synthesis-oriented
AI Analysis

This work addresses the problem of assessing spatial reasoning in MLLMs for applications like robotics and autonomous systems, but it is incremental as it primarily provides a new benchmark.

The paper introduced LEGO-Puzzles, a benchmark with 1,100 visual question-answering samples to evaluate multimodal large language models (MLLMs) on multi-step spatial reasoning, finding that even top MLLMs achieve only about 50% accuracy compared to humans' over 90%.

Multi-step spatial reasoning entails understanding and reasoning about spatial relationships across multiple sequential steps, which is crucial for tackling complex real-world applications, such as robotic manipulation, autonomous navigation, and automated assembly. To assess how well current Multimodal Large Language Models (MLLMs) have acquired this fundamental capability, we introduce LEGO-Puzzles, a scalable benchmark designed to evaluate both spatial understanding and sequential reasoning in MLLMs through LEGO-based tasks. LEGO-Puzzles consists of 1,100 carefully curated visual question-answering (VQA) samples spanning 11 distinct tasks, ranging from basic spatial understanding to complex multi-step reasoning. Based on LEGO-Puzzles, we conduct a comprehensive evaluation of 20 state-of-the-art MLLMs and uncover significant limitations in their spatial reasoning capabilities: even the most powerful MLLMs can answer only about half of the test cases, whereas human participants achieve over 90% accuracy. Furthermore, based on LEGO-Puzzles, we design generation tasks to investigate whether MLLMs can transfer their spatial understanding and reasoning abilities to image generation. Our experiments show that only GPT-4o and Gemini-2.0-Flash exhibit a limited ability to follow these instructions, while other MLLMs either replicate the input image or generate completely irrelevant outputs. Overall, LEGO-Puzzles exposes critical deficiencies in existing MLLMs' spatial understanding and sequential reasoning capabilities, and underscores the need for further advancements in multimodal spatial reasoning.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes