Open Invited Track — 21st IFAC Symposium on System Identification (SysID 2027) : Advances in Identification of Vehicle and Tyre Dynamics for Decision and Control

  • Date: July 7 – 9, 2027
  • Location: ECAM Lyon, Lyon, France

Abstract

Vehicles and their tyres constitute a rich class of nonlinear, uncertain, and safety-critical dynamical systems whose accurate characterisation is central to modern mobility. In line with the theme of SysID 2027—learning models for decision and control—this open invited track focuses on how dynamical models of vehicle and tyre behaviour can be learned from data and then exploited for downstream decision-making. The transition towards electrified, automated, and connected vehicles has considerably increased the demand for reliable models and estimators that operate over a wide range of operating conditions and rely on data collected from onboard, low-cost, and often noisy sensors. The track gathers recent advances in the system identification and estimation of vehicle and tyre dynamics, welcoming contributions on grey-box, black-box, and physics-informed identification, on online and recursive state and parameter estimation (e.g., mass, road friction, slip, sideslip), and on learning-based and hybrid approaches. A central thread is how the resulting models feed decision and control applications—such as motion, stability, and traction control, driver assistance, autonomous driving, safety, comfort, and energy efficiency— placing controller design on the same footing as other end uses of the learned models. Both theoretical developments and experimental or industrial case studies are encouraged. By bringing together researchers from academia and industry, the track fosters exchanges between the system identification community and the automotive community, and highlights open challenges and promising directions for future research.

1 Detailed Description of the Topic

1.1 Motivation and relevance

The automotive sector is undergoing a profound transformation driven by electrification, driving automation, and connectivity. These trends place unprecedented demands on the fidelity of vehicle and tyre models and on the accuracy of the estimation and control algorithms that rely on them. At the vehicle level, accurate models of the chassis, suspension, powertrain, and full-vehicle dynamics underpin motion control, stability and traction control, state estimation, and advanced driver-assistance and autonomous-driving functions. At the tyre level, the sole interface between the vehicle and the road exhibits a strongly nonlinear, temperature- and wear-dependent behaviour that governs handling, comfort, energy consumption, and, ultimately, safety. The two scales are deeply coupled: vehicle behaviour cannot be understood without the tyre–road interaction, and tyre models are only useful once embedded in a vehicle context. This situation makes vehicle and tyre dynamics a particularly fertile and demanding application domain for system identification. It calls for grey-box and physics-informed models that respect the underlying mechanics while remaining tractable, for recursive and adaptive estimators able to track time-varying operating conditions in real time, and for the tight integration of identification and estimation into control and monitoring architectures. The topic is timely because recent advances in machine learning, sensor fusion, and edge computing are reshaping how such models are built and deployed, while industrial actors increasingly require solutions that are certifiable, interpretable, and robust. This track fits squarely within the scope of SysID 2027, whose theme— learning models for decision and control—captures precisely this ambition: the emphasis is on learning dynamical models of vehicle and tyre behaviour, through identification and estimation, that ultimately serve decision and control. Spanning both the vehicle and the tyre scales, it is sufficiently focused to build a coherent session, yet broad enough to attract contributions from the identification, estimation, and control communities alike.

1.2 Scope and topics of interest

The track welcomes theoretical, methodological, experimental, and industrial contributions spanning both the vehicle and the tyre scales. Representative topics include, but are not limited to:

  • Identification of full-vehicle, chassis, suspension, and powertrain dynamics, including lateral, longitudinal, and vertical behaviour.
  • Grey-box, physics-informed, and black-box identification of tyre and tyre–road contact models (e.g., Pacejka-type, brush, and data-driven force models).
  • Linear parameter-varying and nonlinear identification for operating-condition-dependent vehicle and tyre behaviour.
  • Online or offline estimation of vehicle states and physical parameters (mass, inertia, sideslip, yaw rate, road friction, slip, contact forces, load).
  • Sensor fusion and virtual/soft sensing from onboard vehicle sensors, including IMU-, GNSS-, and vision-based approaches.
  • Learning-based and hybrid physics/ML methods, and their interpretability, robustness, and data-efficiency.
  • Integration of identification and estimation into vehicle motion, stability, and traction control, as well as driver-assistance and autonomous-driving functions.
  • Digital twins and condition monitoring of vehicles and tyres for diagnosis and predictive maintenance (e.g., battery/tyre state of health and wear).
  • Experimental validation, benchmark datasets, and industrial case studies from real-world vehicle and tyre data.
1.3 Novelty and expected impact

Vehicle and tyre dynamics offer a uniquely demanding testbed on which learned models meet real, safety-critical constraints. The novelty of this track lies in following a single thread—from learning dynamical models through identification and estimation to their use for decision and control—rather than treating these steps in isolation, and in confronting the methods of the system identification community with the operational realities of modern mobility—limited excitation, harsh and time-varying conditions, low-cost sensing, and stringent requirements on interpretability and certification. Particular attention is given to the interplay between physical modelling and modern data-driven techniques, which is currently one of the most active and debated topics in the field. By gathering academic and industrial perspectives in a single session, the track is expected to surface open problems (e.g., identifiability under limited excitation, uncertainty quantification for certification, transfer across vehicles and road conditions), to encourage the sharing of datasets and benchmarks, and to seed new academia–industry collaborations. It thereby supports the broader mission of the IFAC community to translate identification theory into robust, deployable engineering solutions.

1.4 Target audience and expected contributions

The track targets both academic researchers in system identification, estimation, and control, and industrial practitioners from the automotive and tyre sectors (OEMs, suppliers, and mobilityservice providers). Given the organizers’ complementary academic and industrial affiliations, the track is positioned to attract contributions from both sides and to stimulate exchange between them. We anticipate a session comprising several regular papers, and we plan to actively solicit submissions through the organizers’ networks, including the IFAC Technical Committee 1.1 (Modelling, Identification and Signal Processing), the I-TireLab academic–industrial consortium (University of Poitiers, GIPSA-lab, and Michelin), relevant mailing lists, and direct invitations to leading groups working on vehicle dynamics, tyre modelling, and automotive estimation and control. In line with the conference policy, the organizers commit to promoting the track actively and to securing a sufficient number of high-quality submissions.

2 Information of the Organizers

Organizer 1 (main contact)
  • Name : Hugo Koide
  • Affiliation 1 : Laboratoire d'Informatique et d'Automatique pour les Systèmes (LIAS), France
  • Affiliation 2 : The Michelin Group, France
  • E-mail : hugo.koide@univ-poitiers.fr

Short bio. Hugo Koide was born in London, United Kingdom. He received an M.Eng. degree in mechanical engineering from Imperial College London in 2022, and a Ph.D. degree in automatic control from the University of Poitiers in 2026. He is a member of the LIAS and of the I-TireLab, a joint academic–industrial laboratory involving the University of Poitiers, GIPSA-lab (Grenoble), and Michelin, dedicated to data-driven modeling and control for smart mobility and manufacturing. His work focuses on developing industry-ready digital twin solutions by combining sensor fusion, state estimation, and advanced signal processing techniques to extract actionable insights from real-world vehicle and tyre data. His interests lie at the intersection of control systems, machine learning, physics-based modeling, and intelligent sensing, with a focus on translating cutting-edge research into robust industrial applications.

Organizer 2

Short bio. Guillaume Mercère was born in Cambrai, France, in 1977. He received the M.S. degree in electrical engineering from Caen Engineering School, Caen, France, in 2001, the Ph.D. degree in automatic control from Lille University, Lille, France, in 2004, and the Habilitation ‘a diriger des Recherches from the University of Poitiers, Poitiers, France, in 2012. He has been with the University of Poitiers since 2005, where he has been a Full Professor since September 2024. He is the Director of the I-TireLab, a joint academic–industrial laboratory involving the University of Poitiers, GIPSA-lab (Grenoble), and Michelin, dedicated to data-driven modeling and control for smart mobility and manufacturing. He was Co-Chair of the French Technical Committee on System Identification from 2008 to 2014 and Chair of the IEEE CSS Technical Committee on System Identification and Adaptive Control from 2016 to 2019. He currently chairs the IFAC Technical Committee 1.1 on Modeling, Identification, and Signal Processing. He serves as an Associate Editor for IEEE CSS conferences ACC and CDC and has held visiting appointments at the University of Iceland, Nova Southeastern University (USA), and Politecnico di Milano, Italy. His research interests include model learning and system identification, estimation and optimization theory, subspace-based identification, gray-box and linear parameter-varying models, with applications to vehicle tire–road interaction, heat transfer, flexible manipulators, aeronautics, and image processing.

Organizer 3

Short bio. Hiba Houmsi earned her M.Eng. (Diplôme d’Ingénieur) in electrical engineering from INSA Lyon in 2022, followed by a Ph.D. in control theory from INSA Lyon’s Ampère Laboratory in 2025. She then served as a temporary teaching and research assistant (ATER) at Ecole Centrale de Lyon before joining the Michelin Group as a research engineer, where she now builds industry-ready digital twin solutions by combining model-based design, estimation, sensor fusion, and optimization-based algorithms to turn real-world data into actionable insights. Her research interests span control theory, convex optimization, and embedded real-time systems, particularly where these disciplines meet the practical constraints of real-time, low-cost industrial hardware.