ChatMotion: A Multimodal Multi-Agent for Human Motion Analysis

2025-02-25

AI Summary

Current multimodal large language models (MLLMs) for human motion understanding suffer from a unidirectional instruction-following paradigm, lacking interactivity and dynamic adaptability for multi-perspective analysis. To address this, we propose the first multimodal multi-agent framework tailored for motion analysis, establishing a closed-loop architecture comprising intent-driven reasoning, task decomposition, and modular coordination—thereby introducing the first interactive, multi-perspective motion analysis paradigm. We design MotionCore, a dedicated motion representation module enabling on-demand activation and functional decoupling. Our framework integrates MLLMs, collaborative agent architectures, and modular interface technologies. Evaluated across diverse motion understanding tasks, it achieves accuracy gains of 12.3%–28.7% over state-of-the-art baselines, while significantly improving user engagement and analytical flexibility.

Technology Category

Machine Learning: Large Multimodal Models (LMMs)Computer Vision: Motion & TrackingMultiagent Systems: Multiagent Learning

Application Category

Responsible Web: Machine-in-the-loop, human agency and autonomyUser Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendationSemantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Advancements in Multimodal Large Language Models (MLLMs) have improved human motion understanding. However, these models remain constrained by their"instruct-only"nature, lacking interactivity and adaptability for diverse analytical perspectives. To address these challenges, we introduce ChatMotion, a multimodal multi-agent framework for human motion analysis. ChatMotion dynamically interprets user intent, decomposes complex tasks into meta-tasks, and activates specialized function modules for motion comprehension. It integrates multiple specialized modules, such as the MotionCore, to analyze human motion from various perspectives. Extensive experiments demonstrate ChatMotion's precision, adaptability, and user engagement for human motion understanding.

Problem

• Enhances interactivity in motion analysis
• Decomposes complex tasks into meta-tasks
• Integrates specialized modules for diverse perspectives

Innovation

• Multimodal multi-agent framework
• Dynamic user intent interpretation
• Specialized motion comprehension modules