Generative AI in Embodied Systems: System-Level Analysis of Performance, Efficiency and Scalability

2025-04-26

AI Summary

This work addresses the deployment challenges of generative autonomous agents in embodied AI systems by systematically identifying performance bottlenecks in real-world physical environments: high planning latency, explosive prompt-length growth, and severe degradation in multi-agent collaboration scalability. It presents the first system-level workload characterization of embodied AI agents, introducing a novel four-paradigm classification framework. Guided by benchmark-driven modular modeling and cross-paradigm scalability evaluation, the study identifies seven critical bottlenecks. Building on this analysis, it proposes LLM-aware optimization strategies that significantly improve task success rate, inter-agent communication efficiency, and multi-agent collaborative scalability. The contributions include a rigorous theoretical framework and a systematic methodology for engineering the deployment of embodied generative agents—bridging the gap between foundational LLM capabilities and real-world embodied intelligence.

Technology Category

• Intelligent Robots: Embodied AI
• Multiagent Systems: Multiagent Planning
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Embodied systems, where generative autonomous agents engage with the physical world through integrated perception, cognition, action, and advanced reasoning powered by large language models (LLMs), hold immense potential for addressing complex, long-horizon, multi-objective tasks in real-world environments. However, deploying these systems remains challenging due to prolonged runtime latency, limited scalability, and heightened sensitivity, leading to significant system inefficiencies. In this paper, we aim to understand the workload characteristics of embodied agent systems and explore optimization solutions. We systematically categorize these systems into four paradigms and conduct benchmarking studies to evaluate their task performance and system efficiency across various modules, agent scales, and embodied tasks. Our benchmarking studies uncover critical challenges, such as prolonged planning and communication latency, redundant agent interactions, complex low-level control mechanisms, memory inconsistencies, exploding prompt lengths, sensitivity to self-correction and execution, sharp declines in success rates, and reduced collaboration efficiency as agent numbers increase. Leveraging these profiling insights, we suggest system optimization strategies to improve the performance, efficiency, and scalability of embodied agents across different paradigms. This paper presents the first system-level analysis of embodied AI agents, and explores opportunities for advancing future embodied system design.

Problem

• Analyzing performance and efficiency of embodied AI systems
• Addressing latency and scalability issues in generative agents
• Optimizing system design for multi-agent collaboration

Innovation

• Systematically categorizes embodied systems into four paradigms
• Benchmarks performance across modules, scales, and tasks
• Proposes optimization strategies for efficiency and scalability