Guaranteed Discovery of Control-Endogenous Latent States with Multi-Step Inverse Models

TL;DR

This paper introduces AC-State, an algorithm that guarantees the discovery of minimal, control-relevant latent states from high-dimensional sensory data, enabling effective control and exploration without rewards or demonstrations.

Contribution

The paper proposes a theoretically guaranteed multi-step inverse model with an information bottleneck to identify control-endogenous latent states, advancing state discovery in complex environments.

Findings

Successfully localizes a robot arm with distractions

Explores mazes with multiple agents without rewards

Navigates in complex house simulations effectively

Abstract

In many sequential decision-making tasks, the agent is not able to model the full complexity of the world, which consists of multitudes of relevant and irrelevant information. For example, a person walking along a city street who tries to model all aspects of the world would quickly be overwhelmed by a multitude of shops, cars, and people moving in and out of view, each following their own complex and inscrutable dynamics. Is it possible to turn the agent's firehose of sensory information into a minimal latent state that is both necessary and sufficient for an agent to successfully act in the world? We formulate this question concretely, and propose the Agent Control-Endogenous State Discovery algorithm (AC-State), which has theoretical guarantees and is practically demonstrated to discover the minimal control-endogenous latent state which contains all of the information necessary for controlling the agent, while fully discarding all irrelevant information. This algorithm consists of a multi-step inverse model (predicting actions from distant observations) with an information bottleneck. AC-State enables localization, exploration, and navigation without reward or demonstrations. We demonstrate the discovery of the control-endogenous latent state in three domains: localizing a robot arm with distractions (e.g., changing lighting conditions and background), exploring a maze alongside other agents, and navigating in the Matterport house simulator.