The Sandbox Environment for Generalizable Agent Research (SEGAR)

TL;DR

SEGAR is a customizable, easy-to-use benchmark environment designed to facilitate research on generalization in reinforcement learning by allowing precise control over task distributions and measuring generalization performance.

Contribution

The paper introduces SEGAR, a flexible environment that improves the design, implementation, and evaluation of generalization benchmarks in RL research.

Findings

SEGAR enables easy specification of task distributions.

SEGAR facilitates measuring generalization performance.

Experiments demonstrate SEGAR's utility in various research questions.

Abstract

A broad challenge of research on generalization for sequential decision-making tasks in interactive environments is designing benchmarks that clearly landmark progress. While there has been notable headway, current benchmarks either do not provide suitable exposure nor intuitive control of the underlying factors, are not easy-to-implement, customizable, or extensible, or are computationally expensive to run. We built the Sandbox Environment for Generalizable Agent Research (SEGAR) with all of these things in mind. SEGAR improves the ease and accountability of generalization research in RL, as generalization objectives can be easy designed by specifying task distributions, which in turns allows the researcher to measure the nature of the generalization objective. We present an overview of SEGAR and how it contributes to these goals, as well as experiments that demonstrate a few types of research questions SEGAR can help answer.