Physics-Informed Direct Data-Driven Gain-Scheduled Control for Improving the Extrapolation Performance of Vehicle Lateral Control

This paper proposes a physics-informed direct data-driven control (PI-D3C) framework for designing a gain-scheduled proportional-integral controller for vehicle yaw-rate tracking. Vehicle lateral dynamics vary depending on vehicle velocity, making it difficult for controllers designed using data collected over a limited operating range to maintain satisfactory control performance outside that range. We introduce a novel PI-D3C framework that integrates pure data-driven virtual reference feedback tuning (VRFT) with prior physical knowledge, thereby significantly improving extrapolation performance beyond the operating range covered by the collected data. In the proposed approach, the gain-scheduled controller parameters are designed based on both training data and the yaw-rate response predicted by the physics-based model. To evaluate the effectiveness of the proposed method, numerical simulations are conducted using co-simulation between MATLAB/Simulink and CarSim. The results demonstrate that the controller designed by pure VRFT achieves satisfactory tracking performance within the training velocity region but exhibits significant deterioration in tracking performance outside the training region. In contrast, the proposed controller maintains good yaw-rate tracking performance over a wider velocity range while suppressing excessive variations in the scheduled gains and steering input. These results indicate that incorporating prior physical knowledge into pure data-driven VRFT improves the extrapolation performance of gain-scheduled controllers designed using limited training data.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-11-16
DOI
https://doi.org/10.48550/arxiv.2611.20907
Primary Topic
Vehicle Dynamics and Control Systems
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Physics-Informed Direct Data-Driven Gain-Scheduled Control for Improving the Extrapolation Performance of Vehicle Lateral Control

Shuichi Yahagi, Ansei Yonezawa, Heisei Yonezawa, Itsuro KAJIWARA et al.
arXiv (Cornell University)
Vehicle Dynamics and Control Systems
preprint

Physics-Informed Direct Data-Driven Gain-Scheduled Control for Improving the Extrapolation Performance of Vehicle Lateral Control

Shuichi Yahagi, Ansei Yonezawa, Heisei Yonezawa, Itsuro KAJIWARA, Hiroki Seto, Aoi Koyanagi, Toshiyuki Sugimachi
preprint en

Abstract

This paper proposes a physics-informed direct data-driven control (PI-D3C) framework for designing a gain-scheduled proportional-integral controller for vehicle yaw-rate tracking. Vehicle lateral dynamics vary depending on vehicle velocity, making it difficult for controllers designed using data collected over a limited operating range to maintain satisfactory control performance outside that range. We introduce a novel PI-D3C framework that integrates pure data-driven virtual reference feedback tuning (VRFT) with prior physical knowledge, thereby significantly improving extrapolation performance beyond the operating range covered by the collected data. In the proposed approach, the gain-scheduled controller parameters are designed based on both training data and the yaw-rate response predicted by the physics-based model. To evaluate the effectiveness of the proposed method, numerical simulations are conducted using co-simulation between MATLAB/Simulink and CarSim. The results demonstrate that the controller designed by pure VRFT achieves satisfactory tracking performance within the training velocity region but exhibits significant deterioration in tracking performance outside the training region. In contrast, the proposed controller maintains good yaw-rate tracking performance over a wider velocity range while suppressing excessive variations in the scheduled gains and steering input. These results indicate that incorporating prior physical knowledge into pure data-driven VRFT improves the extrapolation performance of gain-scheduled controllers designed using limited training data.

arXiv (Cornell University)
Vehicle Dynamics and Control Systems
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon