SMAClite: A Lightweight Environment for Multi-Agent Reinforcement Learning

TL;DR

SMAClite is an open-source, lightweight, and flexible environment for multi-agent reinforcement learning based on SMAC, enabling easier experimentation and faster performance compared to the original SMAC environment.

Contribution

It introduces SMAClite, a decoupled and open-source version of SMAC, allowing for easier content creation and faster, more efficient MARL research.

Findings

SMAClite is equivalent to SMAC in training MARL algorithms.

SMAClite outperforms SMAC in runtime speed.

SMAClite uses less memory than SMAC.

Abstract

There is a lack of standard benchmarks for Multi-Agent Reinforcement Learning (MARL) algorithms. The Starcraft Multi-Agent Challenge (SMAC) has been widely used in MARL research, but is built on top of a heavy, closed-source computer game, StarCraft II. Thus, SMAC is computationally expensive and requires knowledge and the use of proprietary tools specific to the game for any meaningful alteration or contribution to the environment. We introduce SMAClite -- a challenge based on SMAC that is both decoupled from Starcraft II and open-source, along with a framework which makes it possible to create new content for SMAClite without any special knowledge. We conduct experiments to show that SMAClite is equivalent to SMAC, by training MARL algorithms on SMAClite and reproducing SMAC results. We then show that SMAClite outperforms SMAC in both runtime speed and memory.