Crawling in Rogue's dungeons with (partitioned) A3C

TL;DR

This paper demonstrates that a partitioned A3C reinforcement learning agent can effectively navigate and descend in procedurally generated rogue-like dungeon environments, achieving high success rates.

Contribution

It introduces a partitioned A3C approach tailored for complex, partially observable, and randomly generated environments like rogue-like games, improving navigation success.

Findings

Agent reaches the stairs in 98% of cases

Partitioned A3C outperforms non-partitioned variants

Effective handling of partial observability and randomness

Abstract

Rogue is a famous dungeon-crawling video-game of the 80ies, the ancestor of its gender. Rogue-like games are known for the necessity to explore partially observable and always different randomly-generated labyrinths, preventing any form of level replay. As such, they serve as a very natural and challenging task for reinforcement learning, requiring the acquisition of complex, non-reactive behaviors involving memory and planning. In this article we show how, exploiting a version of A3C partitioned on different situations, the agent is able to reach the stairs and descend to the next level in 98% of cases.