EmbodiedAgent: A Scalable Hierarchical Approach to Overcome Practical Challenge in Multi-Robot Control

2025-04-14

AI Summary

Addressing critical challenges in multi-robot collaboration—namely task hallucination (generating infeasible instructions) and violation of environmental constraints—this paper proposes a hierarchical embodied agent framework. Methodologically, it introduces: (i) a novel hallucination-resilient mechanism integrating next-action prediction with structured memory for long-horizon task decomposition and dynamic feasibility verification; (ii) MultiPlan+, a large-scale benchmark dataset comprising over 18,000 expert-annotated samples focused specifically on unrealistic tasks; and (iii) RPAS, a comprehensive evaluation framework combining automated metrics with LLM-based expert scoring. The approach unifies hierarchical decision-making, multimodal constraint modeling, and LLM-augmented assessment. Experimental results demonstrate that our method achieves an RPAS score of 71.85%, significantly surpassing state-of-the-art baselines. Furthermore, it successfully enables sustained, cross-platform robotic collaboration for two hours in real-world office environments.

Technology Category

• Intelligent Robots: Multi-Robot Systems
• Multiagent Systems: Multiagent Planning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

This paper introduces EmbodiedAgent, a hierarchical framework for heterogeneous multi-robot control. EmbodiedAgent addresses critical limitations of hallucination in impractical tasks. Our approach integrates a next-action prediction paradigm with a structured memory system to decompose tasks into executable robot skills while dynamically validating actions against environmental constraints. We present MultiPlan+, a dataset of more than 18,000 annotated planning instances spanning 100 scenarios, including a subset of impractical cases to mitigate hallucination. To evaluate performance, we propose the Robot Planning Assessment Schema (RPAS), combining automated metrics with LLM-aided expert grading. Experiments demonstrate EmbodiedAgent's superiority over state-of-the-art models, achieving 71.85% RPAS score. Real-world validation in an office service task highlights its ability to coordinate heterogeneous robots for long-horizon objectives.

Problem

• Addresses hallucination in impractical multi-robot tasks
• Decomposes tasks into executable skills with dynamic validation
• Coordinates heterogeneous robots for long-horizon objectives

Innovation

• Hierarchical framework for multi-robot control
• Next-action prediction with structured memory
• Dynamic validation against environmental constraints