Saved in:
Bibliographic Details
Main Authors: Gu, Xunjiang, Chitta, Kashyap, Golchoubian, Mahsa, Suplin, Vladimir, Gilitschenski, Igor
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.01516
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911640565317632
author Gu, Xunjiang
Chitta, Kashyap
Golchoubian, Mahsa
Suplin, Vladimir
Gilitschenski, Igor
author_facet Gu, Xunjiang
Chitta, Kashyap
Golchoubian, Mahsa
Suplin, Vladimir
Gilitschenski, Igor
contents Robust control policy learning for autonomous driving requires training environments to be both physically realistic and computationally scalable, properties that existing simulators provide only in isolation. We introduce Sim2Sim2Sim, a framework that bridges high-fidelity vehicle simulation and scalable reinforcement learning by distilling simulator dynamics into a highly parallelizable learned dynamics model. By training control policies purely within this distilled environment and deploying them back into the high-fidelity source simulator, we demonstrate more efficient policy optimization and reliable transfer under challenging dynamics. We further show that predictive accuracy alone does not fully characterize a learned dynamics model's suitability as a reinforcement learning training environment, which should also be assessed by the quality of the policies it enables.
format Preprint
id arxiv_https___arxiv_org_abs_2605_01516
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Dynamics Distillation for Efficient and Transferable Control Learning
Gu, Xunjiang
Chitta, Kashyap
Golchoubian, Mahsa
Suplin, Vladimir
Gilitschenski, Igor
Robotics
Robust control policy learning for autonomous driving requires training environments to be both physically realistic and computationally scalable, properties that existing simulators provide only in isolation. We introduce Sim2Sim2Sim, a framework that bridges high-fidelity vehicle simulation and scalable reinforcement learning by distilling simulator dynamics into a highly parallelizable learned dynamics model. By training control policies purely within this distilled environment and deploying them back into the high-fidelity source simulator, we demonstrate more efficient policy optimization and reliable transfer under challenging dynamics. We further show that predictive accuracy alone does not fully characterize a learned dynamics model's suitability as a reinforcement learning training environment, which should also be assessed by the quality of the policies it enables.
title Dynamics Distillation for Efficient and Transferable Control Learning
topic Robotics
url https://arxiv.org/abs/2605.01516