Plan Your Travel and Travel with Your Plan: Wide-Horizon Planning and Evaluation via LLM

2025-06-14

AI Summary

Existing travel planning methods struggle with real-world challenges—including long-horizon planning, multi-constraint satisfaction, personalization, and dynamic temporal dependencies—while lacking evaluation benchmarks that capture itinerary evolution. To address this, we propose the L³ planning paradigm (Long-context, Long-instruction, Long-output) and introduce Multi-dimensional Anticipatory Planning (MAoP) to generate scalable itinerary blueprints. We further present Travel-Sim, the first agent-driven, dynamic travel simulation benchmark that explicitly models event-level temporal dependencies and real-world feasibility constraints. Our approach integrates multi-stage policy decomposition, dynamic constraint-aware planning, and agent-based simulation modeling. On Travel-Sim, our method achieves an average 32.7% improvement in itinerary feasibility over baselines, significantly enhancing rationality, operational feasibility, and alignment with user preferences.

Technology Category

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

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• 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

Abstract

Travel planning is a complex task requiring the integration of diverse real-world information and user preferences. While LLMs show promise, existing methods with long-horizon thinking struggle with handling multifaceted constraints and preferences in the context, leading to suboptimal itineraries. We formulate this as an $L^3$ planning problem, emphasizing long context, long instruction, and long output. To tackle this, we introduce Multiple Aspects of Planning (MAoP), enabling LLMs to conduct wide-horizon thinking to solve complex planning problems. Instead of direct planning, MAoP leverages the strategist to conduct pre-planning from various aspects and provide the planning blueprint for planning models, enabling strong inference-time scalability for better performance. In addition, current benchmarks overlook travel's dynamic nature, where past events impact subsequent journeys, failing to reflect real-world feasibility. To address this, we propose Travel-Sim, an agent-based benchmark assessing plans via real-world travel simulation. This work advances LLM capabilities in complex planning and offers novel insights for evaluating sophisticated scenarios through agent-based simulation.

Problem

• Handling multifaceted constraints in travel planning with LLMs
• Addressing dynamic nature of travel in benchmark evaluations
• Enhancing LLM capabilities for complex itinerary planning

Innovation

• MAoP enables wide-horizon planning via LLMs
• Strategist pre-plans for scalable inference performance
• Travel-Sim benchmark assesses real-world feasibility dynamically