Prioritized Safe Interval Path Planning for Multi-Agent Pathfinding With Continuous Time on 2D Roadmaps

Abstract

We address a challenging multi-agent pathfinding (MAPF) problem for hundreds of agents moving on a 2D roadmap with continuous time. Despite its known potential for producing better solutions compared to typical grid and discrete-time cases, few approaches have been established to solve this problem due to the intractability of collision checks on a large scale. In this work, we propose Prioritized Safe-Interval Path Planning with Continuous-Time Conflicts (PSIPP/CTC) that extends a scalable prioritized planning algorithm to work on the 2D roadmap and continuous-time setup by alleviating intensive collision checks. Our approach involves a novel concept named Continuous-Time Conflict (CTC), which describes a pair among vertices and edges associated with continuous-time intervals within which collisions can happen between agents. We precompute CTCs using geometric neighbor-search and sweeping techniques and annotate roadmaps with the CTCs just once before planning starts. Doing so allows us to efficiently enumerate collision-free time intervals for all vertices and edges and find each agent's path with continous time in prioritized planning. Extensive experimental evaluations demonstrate that PSIPP/CTC significantly outperforms existing methods in terms of planning success rate and runtime while maintaining an acceptable solution quality. As a proof of concept, we also confirmed the effectiveness of the proposed approach on a physics simulation with differential wheeled robots.

A Comprehensive Review on Leveraging Machine Learning for Multi-Agent Path Finding

“…These advancements have not only improved planning capabilities but also expanded the research scope of MAPF in related technological fields. This includes going beyond the simplified assumption of uniform travel time between vertices [30]- [32], extending MAPF to various robotic systems like manipulators [33]- [35], moving beyond the typical one-shot planning problems in MAPF [2], [36], [37], joint planning with target assignments [38]- [41], and executing MAPF solutions with actual robots under timing uncertainties [42]- [45], among others. This evolution broadens the horizons of MAPF technologies, making them increasingly realistic and practical for implementing efficient multi-agent navigation.…”

Periodic Multi-Agent Path Planning

“…Baseline methods. For the FCFS approach, we modify the implementation of Prioritized SIPP by Kasaura, Nishimura, and Yonetani (2022) to our problem setting. For the SO approach, we modify the implementation of CCBS by the authors, which is available online 3 , to our problem setting.…”

“…For each query, planning was considered a failure if the runtime exceeded 1.0 seconds. For PRM, We also use the implementation by Kasaura, Nishimura, and Yonetani (2022). As the parameters for PRM, the number of roadmap vertices is 1000 for the FCFS approach and 20 for the SO approach.…”

Anytime Lifelong Multi-Agent Pathfinding in Topological Maps

“…They solved four MAPF instances in parallel using an anytime strategy called large neighborhood search [22] to plan paths for thousands of agents efficiently. Andreychuk et al [20] and Kasaura et al [33] solved a continuous-time MAPF problem on a topological map. They used a variant of safe interval path planning (SIPP) [34] to handle continuous time conflicts on a graph.…”