Peter Stone's Selected Publications

Classified by TopicClassified by Publication TypeSorted by DateSorted by First Author Last NameClassified by Funding Source


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

Out-of-Distribution Generalization with a SPARC: Racing 100 Unseen Vehicles with a Single Policy.
Bram Grooten, Patrick MacAlpine, Kaushik Subramanian, Peter Stone, and Peter R. Wurman.
In The AAAI Conference on Artificial Intelligence, January 2026.

Download

[PDF]3.9MB  

Abstract

Generalization to unseen environments is a significant challenge in the field ofrobotics and control. In this work, we focus on contextual reinforcementlearning, where agents act within environments with varying contexts, such asself-driving cars or quadrupedal robots that need to operate in differentterrains or weather conditions than they were trained for. We tackle the criticaltask of generalizing to out-of-distribution (OOD) settings, without access toexplicit context information at test time. Recent work has addressed this problemby training a context encoder and a history adaptation module in separate stages.While promising, this two-phase approach is cumbersome to implement and train. Wesimplify the methodology and introduce SPARC: single-phase adaptation for robustcontrol. We test SPARC on varying contexts within the high-fidelity racingsimulator Gran Turismo 7 and wind-perturbed MuJoCo environments, and find that itachieves reliable and robust OOD generalization.

BibTeX Entry

@InProceedings{SPARC_aaai2026,
  author   = {Bram Grooten and Patrick MacAlpine and Kaushik Subramanian and Peter Stone and Peter R. Wurman},
  title    = {Out-of-Distribution Generalization with a SPARC: Racing 100 Unseen Vehicles with a Single Policy},
  booktitle = {The AAAI Conference on Artificial Intelligence},
  year     = {2026},
  month    = {January},
  location = {Singapore, Singapore},
  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.
  },
}

Generated by bib2html.pl (written by Patrick Riley ) on Sat Nov 15, 2025 21:30:11