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

TL;DR

PhysBench is a comprehensive benchmark designed to evaluate and improve vision-language models' understanding of physical phenomena, addressing a key gap in embodied AI capabilities.

Contribution

The paper introduces PhysBench, a large-scale benchmark for physical understanding, and PhysAgent, a framework that enhances VLMs' physical reasoning abilities.

Findings

VLMs excel in common-sense reasoning but struggle with physical understanding.

PhysAgent significantly improves VLMs' performance on physical tasks.

Enhancing physical understanding in VLMs benefits embodied AI applications.

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.