Bridging Planning and Execution: Multi-Agent Path Finding Under Real-World Deadlines

2025-11-26

AI Summary

Multi-agent path finding (MAPF) suffers from a planning-execution gap—conventional models ignore kinodynamic constraints, communication delays, and heterogeneous controller dynamics, leading to plan infeasibility and failure in time-critical tasks. Method: This paper proposes REMAP, the first framework to formally define and solve the MAPF with Realistic Execution Delays (MAPF-RD). Its core innovation is ExecTimeNet, an execution-time-aware neural network that explicitly learns the nonlinear mapping from abstract paths to actual execution durations, and seamlessly integrates into search-based planners (e.g., CBS, MAPF-LNS). The method jointly optimizes over kinodynamic feasibility, communication latency, and controller heterogeneity. Contribution/Results: Evaluated on benchmarks with up to 300 agents, REMAP improves task completion rate by 20% over constant-velocity baselines, significantly enhancing both timeliness and executability of plans in large-scale dynamic environments.

Technology Category

• Multiagent Systems: Multiagent Planning
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

The Multi-Agent Path Finding (MAPF) problem aims to find collision-free paths for multiple agents while optimizing objectives such as the sum of costs or makespan. MAPF has wide applications in domains like automated warehouses, manufacturing systems, and airport logistics. However, most MAPF formulations assume a simplified robot model for planning, which overlooks execution-time factors such as kinodynamic constraints, communication latency, and controller variability. This gap between planning and execution is problematic for time-sensitive applications. To bridge this gap, we propose REMAP, an execution-informed MAPF planning framework that can be combined with leading search-based MAPF planners with minor changes. Our framework integrates the proposed ExecTimeNet to accurately estimate execution time based on planned paths. We demonstrate our method for solving MAPF with Real-world Deadlines (MAPF-RD) problem, where agents must reach their goals before a predefined wall-clock time. We integrate our framework with two popular MAPF methods, MAPF-LNS and CBS. Experiments show that REMAP achieves up to 20% improvement in solution quality over baseline methods (e.g., constant execution speed estimators) on benchmark maps with up to 300 agents.

Problem

• Bridging planning and execution gap in MAPF
• Addressing real-world deadlines for multi-agent pathfinding
• Incorporating execution-time factors into MAPF planning

Innovation

• Execution-informed MAPF planning framework REMAP
• Integrates ExecTimeNet for accurate execution time estimation
• Combines with MAPF-LNS and CBS for deadline-sensitive solutions