Graph augmented Deep Reinforcement Learning in the GameRLand3D environment - Archive ouverte HAL Access content directly
Preprints, Working Papers, ... Year :

Graph augmented Deep Reinforcement Learning in the GameRLand3D environment

(1, 2) , (3) , (3) , (1) , (2) , (3) , (4)
1
2
3
4

Abstract

We address planning and navigation in challenging 3D video games featuring maps with disconnected regions reachable by agents using special actions. In this setting, classical symbolic planners are not applicable or difficult to adapt. We introduce a hybrid technique combining a low level policy trained with reinforcement learning and a graph based high level classical planner. In addition to providing human-interpretable paths, the approach improves the generalization performance of an end-to-end approach in unseen maps, where it achieves a 20% absolute increase in success rate over a recurrent end-to-end agent on a point to point navigation task in yet unseen large-scale maps of size 1km×1km. In an indepth experimental study, we quantify the limitations of end-to-end Deep RL approaches in vast environments and we also introduce "GameRLand3D", a new benchmark and soon to be released environment can generate complex procedural 3D maps for navigation tasks.
Fichier principal
Vignette du fichier
Godot_RL_Agents_2022_AAAI_Workshop (1).pdf (2.29 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03494852 , version 1 (20-12-2021)

Identifiers

  • HAL Id : hal-03494852 , version 1

Cite

Edward Beeching, Maxim Peter, Philippe Marcotte, Jilles Debangoye, Olivier Simonin, et al.. Graph augmented Deep Reinforcement Learning in the GameRLand3D environment. 2021. ⟨hal-03494852⟩
31 View
336 Download

Share

Gmail Facebook Twitter LinkedIn More