Mladen CICIC
I am a maître de conférences (associate professor) at CentraleSupélec, Université Paris-Saclay, in Laboratoire des signaux et systèmes (L2S), Systems and Control group. Previously, I was a postdoctoral fellow at the Department of Civil and Environmental Engineering at University of California, Berkeley, working with Professor Maria Laura Delle Monache, and a postdoctoral researcher at GIPSA-lab, CNRS Grenoble, and Inria Montbonnot, working with Professor Carlos Canudas-de-Wit. I defended my doctoral thesis in March 2021 at KTH Royal Institute of Technology, under supervision of Professor Karl Henrik Johansson. / CentraleSupélec
This H-CODE interdisciplinary workshop aims to bring together researchers working on topics from traffic sensing and estimation, through modeling and prediction, to actuation and control, enabled by Connected and Autonomous Vehicles and similar Intelligent Transporation Systems technologies.
Format: The workshop will be held in person, on 26 November 2026, from 09:00 to 17:00.
Location: CentraleSupélec campus, Gif-sur-Yvette (Southern Paris), room SC.046 (amphi Peugeot) in the Bouygues building.
Registration: Attendance is free, but registration is mandatory before November 12th to allow for an optimal reception of the participants. To register, please fill out this form
Program: The tentative scientific program can be below.
For any inquiries, please contact the organizer, Mladen Čičić (mladen.cicic@centralesupelec.fr)
This workshop is funded by a H-CODE (Human in the loop for control and decision) Interdisciplinary Object of Université Paris-Saclay, grant HC 2026-08.
The program is tentative and the titles, speakers, and order of talks may change.
Abstracts:
Mostafa Ameli – AI-Enabled Mobility Digital Twins for Adaptive Urban Network Management
Managing increasingly complex urban transportation systems requires models that can connect heterogeneous mobility data, individual travel decisions, network dynamics, and operational interventions. This talk presents our ongoing work on developing city-scale mobility digital twins that combine data-driven methods, simulation, artificial intelligence, and optimization to support adaptive transportation planning and control. The framework represents heterogeneous travelers, multimodal choices, departure-time adaptation, route choice, and network congestion as interacting components of the urban mobility system. Multi-source data are used for calibration and network-state inference. Beyond reproducing existing conditions, the digital twin provides a computational environment for testing what-if scenarios and evaluating adaptive interventions under demand changes and disruptions. Particular attention is given to how AI and optimization can support dynamic traffic management, resource allocation, and rerouting strategies while accounting for network performance and resilience. The talk will discuss methodological developments, current applications, and remaining challenges toward more predictive and decision-oriented mobility digital twins.
Negin Alisoltani – Shared Mobility Revolution: Ride-Sharing for Sustainable Urban Solutions
Urban transportation systems are increasingly challenged by traffic congestion, operational inefficiencies, and environmental impacts, primarily resulting from the extensive use of low-occupancy personal vehicles. A promising solution lies in shared mobility systems, particularly ride-sharing, which can optimize trip sharing and reduce traffic congestion. Dr. Alisoltani presents an assessment of the effectiveness of dynamic ride-sharing in alleviating congestion, with a focus on the concept of “shareability,” which quantifies the potential for trips to be shared based on spatial and temporal alignment. The seminar will discuss how shareability can be used to design scalable matching algorithms, evaluate the congestion impacts of ride-sharing under dynamic traffic conditions, and develop fair pricing strategies that encourage passengers to share rides while supporting efficient system operation.
Hossein Nick Zinat Matin – Regularity of the Time-to-Collision Map and Convergence of Learned Hamilton–Jacobi Solutions
Controllers for autonomous vehicles that depend solely on inter-vehicle spacing can behave overly aggressively by ignoring relative velocity. Time-to-collision (TTC) addresses this, but classical kinematic TTC assumes a frozen system state, making it non-predictive and unreliable as feedback. We introduce a dynamically-informed TTC, defined as the hitting time of the collision set under an optimal-velocity prediction flow with merely measurable ($L^\infty$) leader acceleration. This makes the collision estimate consistent with the dynamics, at the cost of a circularity dynamical model. We establish well-posedness and collision avoidance of the resulting model through barrier-domain and near-collision boundary analysis. The analytical core is a regularity theory for TTC on the domain of initial data. We prove That TTC map is in fact $C^1$, on compact subsets, with an explicit gradient. All constants are governed by a single uniform transversality margin that degenerates as tangential collisions are approached. Since the dynamic TTC admits no closed form, online simulation introduces delay that undermines safety guarantees. Recasting the TTC map as value function, we prove a two-sided convergence result for a physics-informed formulation; the $L^2$ error of the learned value function is bounded by the square root of the total loss in HJB framework, and at zero residual the approximation coincides pointwise with the viscosity solution. A two-stage architecture enables delay-free platoon operation, and simulations on a six-vehicle platoon confirm stability across diverse scenarios.
Alessio Iovine – Beyond the Car Ahead: Macroscopic Information for String-Stable Platoons
Stop-and-go waves emerge when small perturbations are amplified through vehicle-to-vehicle interactions. While string stability suppresses this amplification, classical automated platoon designs (when they are really implemented) incur significant trade-offs: predecessor-only feedback with constant spacing requires broadcasting the leader’s state to all vehicles, whereas time-headway policies allow inter-vehicle distances to grow with speed, reducing traffic throughput.
In contrast, human drivers perceive traffic holistically, reacting not just to the immediate predecessor but to broader traffic conditions. This talk demonstrates how automated vehicles can emulate this behavior by combining microscopic feedback from their immediate predecessor with a low-dimensional aggregate of upstream traffic, provided via vehicle-to-infrastructure (V2I) communication. Using trajectory-based analysis and small-gain conditions, we prove asymptotic and disturbance string stability, even under constant spacing and without leader-state broadcasting.
Sheida Nozari – From Microscopic Interactions to Macroscopic Flow: Multi-Scale Control of Mixed-Autonomy Traffic
Traffic congestion emerges from local vehicle interactions but develops into collective traffic waves at the road scale. Addressing this in mixed-autonomy environments, where connected and automated vehicles (CAVs) coexist with human-driven vehicles (HDVs), requires a continuous two-way feedback loop. A multi-scale approach examines how microscopic vehicle dynamics and macroscopic traffic-flow models can be coupled so that each level contributes complementary information, where local interactions describe individual vehicle responses, while macroscopic traffic states capture the evolution and propagation of disturbances across the traffic flow. However, this requires repeated computation of large-scale, interacting multi-agent traffic dynamics, which is computationally demanding for real-time control. To overcome this, physics-informed surrogate models are developed to reproduce macroscopic traffic dynamics at a cost compatible with real-time use. These predictions are incorporated into decentralized CAV control, enabling automated vehicles to anticipate and attenuate stop-and-go waves. The overall aim is to improve global traffic flow while preserving stable CAV–HDV interactions
Gustav Nilsson – Optimal joint control of freeway network actuators and CAVs through a convex relaxation approach
The freeway network control problem, i.e., alleviating congestion by controlling ramp meters and variable speed limits, has been studied for decades. With the introduction of Connected and Autonomous Vehicles (CAVs), which can serve as additional actuators in the system, it is possible to develop control solutions that jointly control CAV speed limits with variable speed limits and ramp meters for regular vehicles. In this talk, a dynamical multi-commodity network flow model based on the cell transmission model will be presented. Although optimal control of such a model is a non-convex optimization problem, one can obtain an exact convex relaxation. Furthermore, some heuristics to make the implementation of the optimal control actions more feasible from a practical viewpoint will be discussed.
