Object-Centric World Models Meet Monte Carlo Tree Search

TL;DR

ObjectZero introduces an object-centric world modeling approach using Graph Neural Networks within a model-based RL framework, effectively enabling planning with Monte Carlo Tree Search in complex environments.

Contribution

The paper presents a novel object-centric RL algorithm that integrates GNN-based world models with MCTS for improved environment understanding and planning.

Findings

Effective learning of object dynamics in complex environments

Successful integration of object-centric models with MCTS for planning

Enhanced environment prediction accuracy using structured representations

Abstract

In this paper, we introduce ObjectZero, a novel reinforcement learning (RL) algorithm that leverages the power of object-level representations to model dynamic environments more effectively. Unlike traditional approaches that process the world as a single undifferentiated input, our method employs Graph Neural Networks (GNNs) to capture intricate interactions among multiple objects. These objects, which can be manipulated and interact with each other, serve as the foundation for our model's understanding of the environment. We trained the algorithm in a complex setting teeming with diverse, interactive objects, demonstrating its ability to effectively learn and predict object dynamics. Our results highlight that a structured world model operating on object-centric representations can be successfully integrated into a model-based RL algorithm utilizing Monte Carlo Tree Search as a planning module.