ATLAS: Constraints-Aware Multi-Agent Collaboration for Real-World Travel Planning

2025-09-29

AI Summary

Real-world travel planning involves explicit, implicit, and dynamically evolving constraints, posing significant challenges for existing LLM-based approaches to simultaneously achieve optimality and practicality. To address this, we propose the first quantitatively effective multi-agent collaborative framework specifically designed for realistic travel planning. Our method innovatively integrates dynamic constraint management, iterative plan critique, and adaptive interleaved search to construct a constraint-aware planning architecture. The system unifies LLM-driven multi-agent coordination, dynamic tool invocation, real-time information retrieval, and multi-turn feedback refinement. Evaluated on the TravelPlanner benchmark, our framework achieves a final success rate of 44.4%, up from 23.3% with prior methods; in real-world deployment, it attains an 84% success rate—substantially outperforming ReAct (59%) and single-agent baselines (27%). This work establishes a new state of the art in operationalizable, constraint-resilient travel planning.

Technology Category

• Multiagent Systems: Multiagent Planning
• Planning, Routing, and Scheduling: Planning with Language Models
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• 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

While Large Language Models (LLMs) have shown remarkable advancements in reasoning and tool use, they often fail to generate optimal, grounded solutions under complex constraints. Real-world travel planning exemplifies these challenges, evaluating agents' abilities to handle constraints that are explicit, implicit, and even evolving based on interactions with dynamic environments and user needs. In this paper, we present ATLAS, a general multi-agent framework designed to effectively handle such complex nature of constraints awareness in real-world travel planning tasks. ATLAS introduces a principled approach to address the fundamental challenges of constraint-aware planning through dedicated mechanisms for dynamic constraint management, iterative plan critique, and adaptive interleaved search. ATLAS demonstrates state-of-the-art performance on the TravelPlanner benchmark, improving the final pass rate from 23.3% to 44.4% over its best alternative. More importantly, our work is the first to demonstrate quantitative effectiveness on real-world travel planning tasks with live information search and multi-turn feedback. In this realistic setting, ATLAS showcases its superior overall planning performance, achieving an 84% final pass rate which significantly outperforms baselines including ReAct (59%) and a monolithic agent (27%).

Problem

• Addresses complex constraints in real-world travel planning
• Improves multi-agent collaboration for dynamic constraint management
• Enhances plan quality through iterative critique and adaptation

Innovation

• Dynamic constraint management for evolving requirements
• Iterative plan critique to refine travel solutions
• Adaptive interleaved search with live information