Out-of-Distribution Generalization with a SPARC: Racing 100 Unseen Vehicles with a Single Policy

TL;DR

This paper introduces SPARC, a single-phase adaptation method for robust out-of-distribution generalization in reinforcement learning, demonstrated on racing and robotics environments without explicit context at test time.

Contribution

We propose SPARC, a simplified, single-phase approach for OOD generalization in contextual reinforcement learning, eliminating the need for multi-stage training.

Findings

SPARC achieves reliable OOD generalization in high-fidelity racing simulations.

SPARC performs well in wind-perturbed MuJoCo environments.

The method simplifies training compared to previous multi-stage approaches.

Abstract

Generalization to unseen environments is a significant challenge in the field of robotics and control. In this work, we focus on contextual reinforcement learning, where agents act within environments with varying contexts, such as self-driving cars or quadrupedal robots that need to operate in different terrains or weather conditions than they were trained for. We tackle the critical task of generalizing to out-of-distribution (OOD) settings, without access to explicit context information at test time. Recent work has addressed this problem by training a context encoder and a history adaptation module in separate stages. While promising, this two-phase approach is cumbersome to implement and train. We simplify the methodology and introduce SPARC: single-phase adaptation for robust control. We test SPARC on varying contexts within the high-fidelity racing simulator Gran Turismo 7 and wind-perturbed MuJoCo environments, and find that it achieves reliable and robust OOD generalization.