Searching with Consistent Prioritization for Multi-Agent Path Finding

Ma, Hang; Harabor, Daniel; Stuckey, Peter J.; Li, Jiaoyang; Koenig, Sven

doi:10.1609/aaai.v33i01.33017643

Cited by 152 publications

(136 citation statements)

References 5 publications

Supporting

Mentioning

136

Contrasting

Order By: Relevance

“…MAPF is an NP-hard problem even when approximating optimal solutions [14], [15]. MAPF planners can be broadly classified into three categories: coupled, decoupled, and dynamically-coupled approaches.…”

Section: A Multi-agent Path Finding (Mapf)mentioning

confidence: 99%

PRIMAL: Pathfinding via Reinforcement and Imitation Multi-Agent Learning

Sartoretti

Kerr

Shi

et al. 2019

IEEE Robot. Autom. Lett.

Self Cite

267

180

View full text Add to dashboard Cite

Multi-agent path finding (MAPF) is an essential component of many large-scale, real-world robot deployments, from aerial swarms to warehouse automation. However, despite the community's continued efforts, most state-of-the-art MAPF planners still rely on centralized planning and scale poorly past a few hundred agents. Such planning approaches are maladapted to real-world deployments, where noise and uncertainty often require paths be recomputed online, which is impossible when planning times are in seconds to minutes. We present PRIMAL, a novel framework for MAPF that combines reinforcement and imitation learning to teach fully-decentralized policies, where agents reactively plan paths online in a partially-observable world while exhibiting implicit coordination. This framework extends our previous work on distributed learning of collaborative policies by introducing demonstrations of an expert MAPF planner during training, as well as careful reward shaping and environment sampling. Once learned, the resulting policy can be copied onto any number of agents and naturally scales to different team sizes and world dimensions. We present results on randomized worlds with up to 1024 agents and compare success rates against state-of-theart MAPF planners. Finally, we experimentally validate the learned policies in a hybrid simulation of a factory mockup, involving both real-world and simulated robots.

show abstract

Section: A Multi-agent Path Finding (Mapf)mentioning

confidence: 99%

PRIMAL: Pathfinding via Reinforcement and Imitation Multi-Agent Learning

Sartoretti

Kerr

Shi

et al. 2019

IEEE Robot. Autom. Lett.

Self Cite

267

180

View full text Add to dashboard Cite

show abstract

“…Prioritised solvers plan paths for each agent individually and avoid collisions with higher priority agents. The priority order can be determined before planning as in Cooperative A* (CA) [8], or determined on the fly as in Priority Based Search (PBS) [24]. Rule-base solvers like Parallel Push and Swap [25] guarantee to find solutions to MAPF in polynomial time, but the quality of these solutions is far from optimal.…”

Section: Multi-agent Path Findingmentioning

confidence: 99%

“…As a result, the overall priority order for path planning is decided by the task assignment sequence. It is worth noting that the path planning part of Algorithm 1 might be incomplete as the prioritised planning is known to be incomplete [24].…”

Section: Algorithm 1 Simultaneous Task Assignment and Path Planningmentioning

confidence: 99%

Integrated Task Assignment and Path Planning for Capacitated Multi-Agent Pickup and Delivery

Chen

Alonso–Mora

Bai

et al. 2021

IEEE Robot. Autom. Lett.

Self Cite

109

View full text Add to dashboard Cite

Multi-agent Pickup and Delivery (MAPD) is a challenging industrial problem where a team of robots is tasked with transporting a set of tasks, each from an initial location and each to a specified target location. Appearing in the context of automated warehouse logistics and automated mail sortation, MAPD requires first deciding which robot is assigned what task (i.e., Task Assignment or TA) followed by a subsequent coordination problem where each robot must be assigned collision-free paths so as to successfully complete its assignment (i.e., Multi-Agent Path Finding or MAPF). Leading methods in this area solve MAPD sequentially: first assigning tasks, then assigning paths. In this work we propose a new coupled method where task assignment choices are informed by actual delivery costs instead of by lower-bound estimates. The main ingredients of our approach are a marginal-cost assignment heuristic and a metaheuristic improvement strategy based on Large Neighbourhood Search. As a further contribution, we also consider a variant of the MAPD problem where each robot can carry multiple tasks instead of just one. Numerical simulations show that our approach yields efficient and timely solutions and we report significant improvement compared with other recent methods from the literature.

show abstract

“…This has been done previously through a centralized controller; see Turpin et al (2013a), Chalaki and Malikopoulos (2019). In general, finding an optimal ordering is generally NP-Hard and an optimal ordering is not always guaranteed to exist; see Ma et al (2019). To reduce the complexity of ordering agents, much work has been done to decentralize the ordering problem.…”

Section: Related Workmentioning

confidence: 99%

An energy-optimal framework for assignment and trajectory generation in teams of autonomous agents

Beaver

Malikopoulos

2020

Systems & Control Letters

View full text Add to dashboard Cite

In this paper, we present a decentralized approach to solving the problem of moving N homogeneous agents into N, or more, goal locations along energy-minimizing trajectories. We propose a decentralized framework which only requires knowledge of the goal locations by each agent. The framework includes guarantees on safety through dynamic constraints, and a method to impose a dynamic, global priority ordering on the agents. A solution to the goal assignment and trajectory generation problems are derived in the form of a binary program and a linear system of equations. We also present the conditions for global optimality and characterize the assumptions under which the algorithm is guaranteed to converge to a unique assignment of agents to goals. Finally, we validate the efficacy of our approach through a numerical simulation in MATLAB.

show abstract

Searching with Consistent Prioritization for Multi-Agent Path Finding

Cited by 152 publications

References 5 publications

PRIMAL: Pathfinding via Reinforcement and Imitation Multi-Agent Learning

PRIMAL: Pathfinding via Reinforcement and Imitation Multi-Agent Learning

Integrated Task Assignment and Path Planning for Capacitated Multi-Agent Pickup and Delivery

An energy-optimal framework for assignment and trajectory generation in teams of autonomous agents

Contact Info

Product

Resources

About