POGEMA: Partially Observable Grid Environment for Multiple Agents

TL;DR

POGEMA is a flexible, scalable grid environment designed to benchmark challenging partially observable multi-agent pathfinding problems, facilitating research in planning and learning integration.

Contribution

It introduces a customizable sandbox environment specifically for PO-MAPF, bridging the gap between AI planning and learning methods.

Findings

Provides a tunable environment for PO-MAPF research

Enables testing of planning and learning algorithms in partially observable settings

Supports scalable and flexible multi-agent pathfinding experiments

Abstract

We introduce POGEMA (https://github.com/AIRI-Institute/pogema) a sandbox for challenging partially observable multi-agent pathfinding (PO-MAPF) problems . This is a grid-based environment that was specifically designed to be a flexible, tunable and scalable benchmark. It can be tailored to a variety of PO-MAPF, which can serve as an excellent testing ground for planning and learning methods, and their combination, which will allow us to move towards filling the gap between AI planning and learning.