GROOT: Learning to Follow Instructions by Watching Gameplay Videos

TL;DR

GROOT is a novel controller that learns to follow open-ended instructions in open-world environments by watching gameplay videos, effectively bridging the gap between human and machine performance in complex tasks.

Contribution

It introduces a new learning framework and architecture for instruction following from videos, creating a structured goal space without needing text annotations.

Findings

GROOT closes the human-machine performance gap.

Achieves a 70% win rate over baseline agents.

Induces a goal space with emergent properties like goal composition.

Abstract

We study the problem of building a controller that can follow open-ended instructions in open-world environments. We propose to follow reference videos as instructions, which offer expressive goal specifications while eliminating the need for expensive text-gameplay annotations. A new learning framework is derived to allow learning such instruction-following controllers from gameplay videos while producing a video instruction encoder that induces a structured goal space. We implement our agent GROOT in a simple yet effective encoder-decoder architecture based on causal transformers. We evaluate GROOT against open-world counterparts and human players on a proposed Minecraft SkillForge benchmark. The Elo ratings clearly show that GROOT is closing the human-machine gap as well as exhibiting a 70% winning rate over the best generalist agent baseline. Qualitative analysis of the induced goal space further demonstrates some interesting emergent properties, including the goal composition and complex gameplay behavior synthesis. The project page is available at https://craftjarvis-groot.github.io.