PhysBench: Benchmarking and Enhancing Vision-Language Models for Physical World Understanding

Wei Chow, Jiageng Mao, Boyi Li, Daniel Seita, V. Guizilini, Yue Wang

arXiv:2501.16411 · 2026-07-27 공개 · arXiv · PDF

benchmarking vision-language-models embodied-ai physical-priors physical-dynamics physbench physical-world-understanding physagent

Abstract

Understanding the physical world is a fundamental challenge in embodied AI, critical for enabling agents to perform complex tasks and operate safely in real-world environments. While Vision-Language Models (VLMs) have shown great promise in reasoning and task planning for embodied agents, their ability to comprehend physical phenomena remains extremely limited. To close this gap, we introduce PhysBench, a comprehensive benchmark designed to evaluate VLMs' physical world understanding capability across a diverse set of tasks. PhysBench contains 10,002 entries of interleaved video-image-text data, categorized into four major domains: physical object properties, physical object relationships, physical scene understanding, and physics-based dynamics, further divided into 19 subclasses and 8 distinct capability dimensions. Our extensive experiments, conducted on 75 representative VLMs, reveal that while these models excel in common-sense reasoning, they struggle with understanding the physical world -- likely due to the absence of physical knowledge in their training data and the lack of embedded physical priors. To tackle the shortfall, we introduce PhysAgent, a novel framework that combines the generalization strengths of VLMs with the specialized expertise of vision models, significantly enhancing VLMs' physical understanding across a variety of tasks, including an 18.4\% improvement on GPT-4o. Furthermore, our results demonstrate that enhancing VLMs' physical world understanding capabilities can help embodied agents such as MOKA. We believe that PhysBench and PhysAgent offer valuable insights and contribute to bridging the gap between VLMs and physical world understanding.

한국어 요약

한 줄 요약

PhysBench 벤치마크와 PhysAgent 프레임워크를 제안하여 VLM의 물리 세계 이해 능력을 평가 및 향상시킨다.

핵심 기여도

핵심 아이디어

기존 VLM은 일반 지식 추론 능력은 뛰어나지만, 물리 세계 이해 능력은 제한적이다. 이는 훈련 데이터에 물리 지식이 결여되어 있고, 물리적 사전지식이 내재화되지 않았기 때문으로 분석된다. 이를 해결하기 위해, **PhysAgent**는 VLM의 일반화 능력과 **시각 기초 모델**(vision foundation models)의 전문 지식을 결합하는 새로운 프레임워크를 제안한다. 또한, **물리 지식 메모리**(physics knowledge memory) 모듈을 통해 필요한 물리 지식을 선택적으로 호출할 수 있도록 설계되어, **깊이 추정**(depth estimation), **수치 거리 계산**(numerical distance calculation) 등 VLM이 취약한 작업을 보완한다. 이는 기존의 물리 추론 방법과 달리, **수작업된 처리 로직**(manually predefined processing logic) 없이도 개방형 문제 해결 능력을 유지한다.

기술적 접근법

주요 결과

의의 및 한계

**의의**:

**한계**:

실용적 활용