Automated Mobility in Mixed Traffic

Motivation and Aim: 

The integration of automated vehicles (AVs) into existing transportation systems is transforming our mobility, presenting unprecedented challenges and opportunities in mixed-traffic environments. In mixed traffic, AVs of varying automation levels coexist with and navigate alongside human-driven vehicles (HDVs) and vulnerable road users (e.g., cyclists and pedestrians). While AVs operate through algorithms, human drivers rely on intuition, experience, and social cues, creating intricate differences in decision-making approaches. These new realities will lead to unprecedented road traffic conditions, accompanied by novel types of interactions, which could have significant implications for both traffic safety and efficiency, uncertain and hard to analyze and predict.

Data-driven, empirical, model-based, and simulation-based research with the emerging powerful AI algorithms and tools are considered critical for understanding the complex dynamics, interactive behaviors, and their impact in mixed traffic. However, several challenges still hinder progress, e.g., the generalization capability of the data-driven modeling, discrepancies between simulation and reality, the lack of high-quality mixed-traffic datasets, and the absence of in-depth collaboration between academia and industry.

 Bringing together diverse perspectives and expertise, this workshop aims to mitigate these gaps, advance the understanding of mixed traffic dynamics, and shape the development of AI-driven mobility systems that are both innovative and responsive to societal needs.  Building upon the success and experience of previous versions of the workshops at ITSC 2024 and ITSC 2023, this third edition further pushes forward the research for automated mobility in mixed traffic by: 

• Providing a unique opportunity for knowledge sharing by gathering together notable researchers in the domain and experts from the leading data collection and vehicle automation companies;

• Showcasing the available emerging datasets, their formats, and structure, and discussing their limitations, and challenges for the current research;

• Showcasing and validating state-of-the-art modelling methods and assumptions, with mixed traffic flow datasets; 

• Identifying current research gaps and future research directions, as well as the opportunities for creating synergy between data-driven and theory-driven research;

• Presenting the new IEEE ITSS Technical Committee with its community website for sharing relevant resources (open-sourced datasets, simulation tools and platforms, and pertinent publications).

Participants of this workshop will have the opportunity to communicate with other researchers and experts face-to-face. The goals are to share best practices, discuss common problems that have not been addressed, and gain insights on future research directions, so as to stay ahead of the curve. Additionally, a set of relevant research resources, e.g., open-sourced datasets with detailed summaries, simulation platforms and tools, relevant publication list, will be shared with the participants after the workshop.

Topics of Interest: 

Interested researchers are invited to present their works on the following relevant topics, including but not limited to:

• Automated mobility and mixed traffic related datasets;

• Data collection, processing, managing, and publishing;

• Mixed traffic status prediction (long/medium/short term);

• Behavioural modelling and interaction in mixed traffic;

• Traffic flow in mixed traffic;

• AI in data-driven research for mixed traffic; 

• LLM, VLM, and VLA applied for AVs; 

• Meaningful human control in mixed autonomy;  

• Robustness, transparency, and trustworthiness of AI in mixed traffic; 

• Impact evaluation methods of mixed traffic; 

• Empirical evaluation of different vehicle automation levels; 

• Driving behavioral adaptation in mixed traffic; 

• Energy consumption/demand in mixed traffic; 

• Empirical studies and field tests about mixed autonomy; 

• Assumptions and simulation models for mixed traffic;

• Open-access and reproducibility of research on mixed traffic;

• Policies, regulations, and codes of practice.

Speakers’ Topic and Bio

• Lines, Signs, Traffic Lights, and High-Definition Maps - Government’s Role in Enabling the Deployment of Connected and Automated Vehicles

Amit Trivedi, Assurance & Innovation Lead, Department of Transport and Main Roads (Queensland), Australia

• Abstract: 

The Department of Transport and Main Roads (TMR) is delivering the Cooperative and Highly Automated Driving (CHAD) pilot, exploring the potential impacts of introducing Cooperative and Automated Vehicles (CAVs) on Queensland roads with a focus on physical road infrastructure, digital infrastructure, interaction with other road users and driver behaviour (the human to machine interface).

Automated Vehicles (AVs) are an emerging technology which promises to remove some or all driving tasks from human control. In recent years, technological advances have made AVs increasingly viable, with low levels of automation already in market‐ready vehicles. So far, studies carried out in Australia are predominantly literature reviews and/or observations of products already in the market designed as Automated Driver Assist Systems (ADAS) for a vigilant human driver. Considering two to four years design, test, and verification cycle, these studies were predominantly assessing older technologies. To overcome this limitation, the CHAD pilot sought to assess the limitations of the Artificial Intelligence (AI) which powers detection technology instead of merely observing the results of market-ready products. 

While physical infrastructure, such as lines, signs, and traffic lights, plays an important role, transitioning from semi‐automated to fully automated widespread operation is the current problem facing most key AV developers. Some of these challenges can be alleviated by providing AVs with a prior high-definition (HD) map. Prior HD maps encode road‐level features such as the position of street signs and lane markings, with up to centimetre accuracy. This allows for the vehicle to verify the data received by its sensors against the prior map, and to even “fill in the gaps” where the incoming sensor data is limited due to difficulties such as rain or occlusions.

Conducting a comprehensive literature review into the use of prior maps in AVs, TMR was focused on understanding the potential role for government in developing, monitoring, and maintaining these maps.

This presentation will discuss the work undertaken in Queensland to ready for the deployment of CAVs from a physical infrastructure perspective. It will cover key issues concerning the physical infrastructure needs and use of prior maps, it will consider what overseas Governments have done to assist the development and deployment of prior HD maps, and what data needs to be shared between governments and private enterprises.

• • Speaker’s Bio:

Amit Trivedi leads TMR’s Safety Technology Projects team.  One of Amit’s responsibilities includes assessing the safety benefits and impacts of introducing connected and automated vehicles on Queensland’s roads. Amit has over 25 years’ experience in Systems Engineering, directing and managing risk, assurance, and HSEQ management systems.  His experience extends to the oil and gas, petrochemical, pharmaceutical, manufacturing, and transport sectors; with dual qualifications in Engineering, he is well-positioned to address organisational risks through engineering, regulation, and compliance.

• Mitigating Metropolitan Carbon Emissions with Dynamic Eco-driving at Scale

Cathy Wu, Associate Professor in LIDS, CEE, and IDSS, Massachusetts Institute of Technology, USA

• Abstract: 

The sheer scale and diversity of transportation make it a formidable sector to decarbonize. Here, we consider an emerging opportunity to reduce carbon emissions: the growing adoption of semi-autonomous vehicles, which can be programmed to mitigate stop-and-go traffic through intelligent speed commands and, thus, reduce emissions. But would such dynamic eco-driving move the needle on climate change? A comprehensive impact analysis has been out of reach due to the vast array of traffic scenarios and the complexity of vehicle emissions. We address this challenge with large-scale scenario modeling efforts and by using multi-task deep reinforcement learning with a carefully designed network decomposition strategy. We perform an in-depth prospective impact assessment of dynamic eco-driving at 6,011 signalized intersections across three major US metropolitan cities, simulating a million traffic scenarios. Overall, we find that vehicle trajectories optimized for emissions can cut city-wide intersection carbon emissions by 11-22%, without harming throughput or safety, and with reasonable assumptions, equivalent to the national emissions of Israel and Nigeria, respectively. We find that 10% eco-driving adoption yields 25%-50% of the total reduction, and nearly 70% of the benefits come from 20% of intersections, suggesting near-term implementation pathways. More broadly, this work paves the way for large-scale analysis of traffic externalities, such as time, safety, and air quality, and the potential impact of solution strategies.

• • Speaker’s Bio:

Cathy Wu is the Class of 1954 Career Development Associate Professor at MIT in LIDS, CEE, and IDSS. She holds a Ph.D. from UC Berkeley, and B.S. and M.Eng. from MIT, all in EECS, and completed a Postdoc at Microsoft Research. Her research advances machine learning for control and optimization in mobility. She is broadly interested in AI for Engineering. Cathy has received a number of awards, including the NSF CAREER, PhD dissertation awards, and publications with distinction. She serves on the Board of Governors for the IEEE ITSS, is a Program Co-chair for RLC 2025, and is an Associate Editor (or equivalent) for TR Part C, ICML, NeurIPS, and ICRA. She is also the inaugural Chair and Co-founder of the REproducible Research In Transportation Engineering (RERITE) Working Group.

• Improving Safety in Autonomous Intersections

Hai L. Vu, Deputy Dean Research, Faculty of Engineering, Professor in Transport Engineering, Monash University, Australia

• Abstract: 

To be determined (TBD).

• • Speaker’s Bio:

Hai L. Vu is a Professor of Intelligent Transport Systems (ITS) and Deputy Dean (Research) in the Faculty of Engineering at Monash University, Australia. He has published over 250 scientific papers in journals and conferences, and is a recipient of the 2012 Australian Research Council (ARC) Future Fellowship and the Victoria Fellowship Award for his research and leadership in ITS. His research interests include modelling and design of complex networks, stochastic optimization, and control with applications to connected autonomous vehicles, transport network planning, and mobility management.

• Driving Behavior and Risk Prediction for Mixed Human-Machine Autonomy towards Harmonious Roadsharing 

Junmin Wang, Professor, Fletcher Stuckey Pratt Chair in Engineering, The University of Texas at Austin, USA

• Abstract: 

The rapid introduction of advanced ground vehicle technologies has created increasingly diverse traffic environments, characterized by heterogeneity in vehicle autonomy levels, manufacturers, and user characteristics. Vehicle autonomy spans human-driven, human-machine shared control, and fully autonomous systems, each exhibiting distinct yet interacting decision-making patterns. Even within a given autonomy level, manufacturer-specific control policies and human variability generate substantial diversity in driving behavior. Achieving not only safe but also harmonious road sharing under such heterogeneity requires predictive intelligence capable of modeling both driving behavior and human risk perception.  In particular, the wide variability in perceived safety and risk necessitates equipping autonomous agents with the ability to infer and adapt to surrounding road participants’ latent risk sensitivities.

This talk presents some recent advances in predictive modeling of driving behavior and risk perception for mixed human-machine autonomy. We discuss frameworks that integrate behavioral prediction, risk inference, and interaction-aware decision-making, enabling safe, efficient, and socially compatible coordination in mixed traffic.  These results point toward scalable approaches for harmonizing heterogeneous road users and accelerating the adoption of autonomous technologies in real-world driving environments.

• • Speaker’s Bio:

Junmin Wang is the Fletcher Stuckey Pratt Chair in Engineering and a Professor in Mechanical Engineering at the University of Texas at Austin.  In 2008, he started his academic career at Ohio State University, where he founded the Vehicle Systems and Control Laboratory, was early promoted to Associate Professor in September 2013, and then very early promoted to Full Professor in June 2016.  In 2018, he left Ohio State University and joined the University of Texas at Austin as the Accenture Endowed Professor in Mechanical Engineering.  Professor Wang has a wide range of research interests covering control, modeling, estimation, optimization, diagnosis, and AI for dynamical systems, especially for automotive, vehicle, transportation, mobility, robotic, energy storage, and manufacturing applications.  Prof. Wang’s research contributions include the development of control and estimation methods that advance efficiency, driving safety, and emission performance of conventional, electrified, connected and autonomous/automated vehicles. He has five years of full-time industrial research experience (2003-2008) at Southwest Research Institute (San Antonio, Texas) where he was a Senior Research Engineer and led research projects sponsored by more than 50 industrial companies and governmental agencies worldwide. 

Professor Wang is the author or co-author of more than 400 peer-reviewed publications (8 of which received best paper awards from IEEE, ASME, and SAE), including 200 journal articles and 13 U.S. and European patents. Prof. Wang is a recipient of the ASME Charles Stark Draper Innovative Practice Award, IEEE Best Vehicular Electronics Paper Award, IEEE Andrew P. Sage Best Transactions Paper Award, IEEE Transactions on Fuzzy Systems Outstanding Paper Award, Ohio State University Lumley Interdisciplinary Research Award, National Science Foundation (NSF) CAREER Award, SAE International Vincent Bendix Automotive Electronics Engineering Award, and Office of Naval Research Young Investigator Program (ONR-YIP) Award. Prof. Wang is an IEEE Vehicular Technology Society Distinguished Lecturer, IEEE Systems Man and Cybernetics Society Distinguished Lecturer, IEEE Industrial Electronics Society Distinguished Lecturer, Fulbright Distinguished Scholar, SAE Fellow, ASME Fellow, AAIA Fellow, and IEEE Fellow.

Professor Wang serves as a Senior Editor, Editor, Technical Editor, or Associate Editor for the IEEE Vehicular Technology Magazine, IEEE Transactions on Vehicular Technology, IEEE/ASME Transactions on Mechatronics, IFAC Control Engineering Practice, IFAC Mechatronics, ASME Transactions Journal of Dynamic Systems, Measurement and Control, Journal of the Franklin Institute, and SAE International Journal of Engines. He has served as Chair of the ASME Automotive and Transportation Systems Technical Committee, Chair of the SAE International Control and Calibration Committee, member (Liaison for IEEE Control Systems Society) of the IEEE Transportation Electrification Steering Committee, and Vice Chair of the IFAC Technical Committee on Mechatronic Systems.

• Autonomous Vehicle Take-over Strategy Using Preference-based Reinforcement Learning

Lishengsa Yue, Associate Professor, College of Transportation Engineering, Tongji University, China

• Abstract: 

Takeover request (TOR) policy plays a critical role in ensuring the safety of autonomous driving. However, existing approaches often rely on static rules or fixed thresholds, which fail to account for drivers’ cognitive states and inter-group differences, thereby limiting safety performance. Preference-based Reinforcement Learning (PbRL) offers a promising alternative by modeling driver-environment-performance relations as latent preferences, facilitating personalized TOR alignment. Such preferences are only reflected indirectly through performance, making learning inherently difficult. PbRL is also sensitive to random seeds, leading to unstable results. Moreover, without a priori policy ranking, identifying optimal policy requires extensive empirical testing, which is inefficient in practice. To address these limitations, we propose a Double-Layer Preference-based Reinforcement Learning (DL-PbRL) framework that incorporates preference modeling at both trajectory and step levels. A reward model is trained to capture the preferences of diverse driver groups, classified by age and driving experience, enabling dynamic alignment between TOR intensity and driver cognitive state. The framework introduces a joint loss function across both layers, improving preference extraction and reducing sensitivity to initialization. A two-phase driving simulator experiment compares DL-PbRL against static and dynamic baselines. Results show that young drivers adapt better to high TOR intensity, while experienced drivers respond effectively to lower-intensity TOR. The proposed method reduces takeover time by about 13.79% compared to real-world baseline strategies. Ablation studies further validate the contribution of the joint loss in decreasing seed sensitivity evaluated by variance, CV, RMAD, and MAG. This work provides a novel cognitive state-aware TOR design framework, offering a robust and efficient solution to enhance shared control safety in autonomous driving.

• • Speaker’s Bio:

Lishengsa Yue is an Associate Professor at Tongji University. He holds dual Ph.D. degrees from the University of Central Florida and Tongji University. His main research interests include human–machine interaction and collaboration in intelligent systems. He has received fundings from the U.S. Department of Transportation and the National Natural Science Foundation of China. Yue has published more than 60 papers in prestigious journals and conferences such as ICCV, IEEE TITS, and TR-Part F. He is a recipient of the Chinese Government Award for Outstanding Students Abroad and the Shanghai Overseas Leading Talent Program. Dr. Yue serves as an editor for the Transportation Research Record, and he also serves on the CAA Technical Committees on Vehicle Control and Intelligent Vehicle.

• Shaping Mixed-Autonomy Networks: A Framework for Dynamic Routing and Pricing

Meead Saberi, Associate Professor, School of Civil and Environmental Engineering, University of New South Wales, Australia

• Abstract: 

The emergence of connected and autonomous vehicles (CAVs) alongside human-driven vehicles (HVs) presents new challenges and opportunities for urban traffic management. While CAVs can follow centrally coordinated, system-optimal (SO) routes, HVs typically select user-equilibrium (UE) paths based on individual preferences, often increasing congestion. This study presents a dynamic framework for integrated routing, mode choice, and pricing in mixed-autonomy networks. Using an extended three-dimensional network fundamental diagram (3D-NFD), the model captures and regulates bimodal traffic flows. A tri-level structure is developed that includes (1) a simulation-based dynamic traffic assignment model to generate realistic equilibrium flows; (2) a feedback controller that computes dynamic, location-specific tolls to incentivize efficient travel behavior for HVs and SO routing for CAVs; and (3) a nested logit mode choice model to estimate modal shifts under elastic demand. Applied to a network model of Melbourne, the framework demonstrates that targeted pricing and CAV routing control can significantly reduce congestion and improve system efficiency, offering a practical path toward sustainable urban mobility in mixed-autonomy contexts.

• • Speaker’s Bio:

Meead Saberi is an Associate Professor in the School of Civil and Environmental Engineering at the University of New South Wales (UNSW), Sydney, Australia. Prior to joining UNSW, he was an academic at Monash University, Melbourne, from 2014 to 2018. He holds a PhD degree in Transportation Systems Analysis and Planning from Northwestern University, USA. He also has a Master’s degree in Transportation Engineering from Portland State University, USA, and a Bachelor’s degree in Civil Engineering from Ferdowsi University of Mashhad, IRAN. He is leading the CityX research lab as part of the Research Centre for Integrated Transport Innovation (rCITI), which focuses on scientific understanding of cities through modelling, simulation, data analytics, and visualisation. His research interests and experience cover a range of transportation engineering areas, including traffic flow theory & characteristics, large-scale transportation network modelling, complex networks, pedestrian crowd dynamics and simulation, and urban data analytics & visualisation.

Dr. Saberi is also a co-founder of footpath.ai, a UNSW spinout that scales and automates mapping of the walking infrastructure with GeoAI and computer vision.

• Autonomous Vehicle Systems for Transportation in Controlled Environments

Nan Zheng, Associate Professor, Department of Civil and Environmental Engineering, Monash University, Australia

• Abstract: 

This presentation will give an overview over a series of R&D project which develops technologies for operating unmanned transportation at open-cut (or open-pit) mining sites. The developments leverage novel modeling, analytical, and computational methods for road infrastructure planning, real-time monitoring, fault detection and prevention, movement and control, operation scheduling, and digital-twin and visualization.

• • Speaker’s Bio:

Nan Zheng is an Associate Professor of Transport Engineering at Monash University. He leads the MARRS research center and has over 15 years of research and development in “transport system modeling and control”, publishing over 100 impactful articles on the related area, including autonomous vehicles, intelligent transportation, and system optimization and control. Nan’s team has been working on autonomous transportation solutions for mining operation applications and commercialization overseas.

• Connected Vehicle-based Adaptive Signal Control: A Novel Stochastic Framework

Shaocheng Jia, Research Fellow, National University of Singapore, Singapore

• Abstract: 

Optimizing traffic signal control is crucial for improving efficiency in congested urban environments. Current adaptive signal control systems predominantly rely on on-road detectors, which entail significant capital and maintenance costs, thereby hindering widespread implementation. In this talk, a novel connected vehicle (CV)-based adaptive signal control (CVASC) framework is presented that optimizes signal plans on a cycle-by-cycle basis without the need for on-road detectors, leveraging partial CV data. The framework comprises a consequential system delay (CSD) model, deterministic penetration rate control (DPRC), and stochastic penetration rate control (SPRC). The CSD model analytically estimates vehicle arrival rates and, consequently, the total junction delay, utilizing CV penetration rates as essential inputs. Employing the CSD model without considering CV penetration rate uncertainty results in fixed vehicle arrival rates and leads to DPRC. On the other hand, incorporating CV penetration rate uncertainty accounts for uncertain vehicle arrival rates, establishing SPRC, which poses a high-dimensional, non-convex, and stochastic optimization problem. An analytical stochastic delay model using generalized polynomial chaos expansion is proposed to efficiently and accurately estimate the mean, variance, and their gradients for the CSD model within SPRC. To solve DPRC and SPRC, a gradient-guided golden section search algorithm is introduced. Comprehensive numerical experiments and VISSIM simulations demonstrate the effectiveness of the CVASC framework, emphasizing the importance of accounting for CV penetration rate uncertainty and uncertain vehicle arrival rates in achieving optimal solutions for adaptive signal optimizations.

• • Speaker’s Bio:

Shaocheng Jia is currently a Research Fellow at the National University of Singapore, and a member of IEEE, IEEE ITSS, INFORMS, INFORMS TSL, CHTS, and IACIP. He received his B.Eng. degree from the Department of Electronic Information Engineering, China University of Petroleum (Beijing), M.Eng. degree from the Department of Automation, Tsinghua University, and Ph.D. degree from the Department of Civil Engineering, the University of Hong Kong, in 2018, 2021, and 2025, respectively. His research interests include connected and automated transportation, intelligent perception and control, stochastic modeling and optimization, and artificial intelligence. Dr. Jia has published more than 20 papers in prestigious journals and conferences, including Transportation Science, TR-Part B, TR-Part C, IEEE TITS, ISTTT, IEEE ITSC, etc., and 6 national invention patents. He is a recipient of the Outstanding Master’s Dissertation Awards from CHTS and Tsinghua University. He has also actively participated in various professional events by serving as session chairs, workshop organizers, and reviewers for dozens of journals and conferences.

• Automated Mobility: Opportunities, Challenges, and Outlooks

Xuan (Sharon) Di, Associate Professor, Department of Civil Engineering and Engineering Mechanics, Columbia University, USA

• Abstract: 

TBD.

• • Speaker’s Bio:

Xuan (Sharon) Di is an Associate Professor in the Department of Civil Engineering and Engineering Mechanics at Columbia University in the City of New York, and co-chairs the Smart Cities Center in the Data Science Institute. Dr. Di directs the DitecT (Data and innovative technology-driven Transportation) Lab, focusing on transportation systems. Her overarching research mission is to empower mobility for all, emphasizing the use of technology for social good. She is currently focused on pioneering the development of digital twins for urban transportation management, leveraging cyber-physical systems technology. Within this framework, her research spans diverse areas, including multi-modal mobility optimization, autonomous vehicle control on shared roads with humans, and the intersection of transportation with health considerations..

• [Topic TBD]

Jiaqi Ma, Professor at the Samueli School of Engineering and Director of the FHWA/UCLA Center of Excellence on New Mobility and AVs, UCLA, USA

• Abstract: 

TBD.

• • Speaker’s Bio:

Jiaqi Ma is the Director of the FHWA/UCLA Center of Excellence on New Mobility and Automated Vehicles, a Professor at the UCLA Samueli School of Engineering, Director of the UCLA Mobility Lab, and Associate Director of the UCLA Institute of Transportation Studies. He has led and managed numerous research projects funded by the U.S. Department of Transportation, National Science Foundation, state Departments of Transportation, and other federal, state, and local agencies. His primary research interests include automated driving, mobile robotics, multimodal sensing, cooperative perception and decision-making, mobility digital twins, spatial data mining, reasoning, and simulation. Dr. Ma serves as Editor-in-Chief of the IEEE Open Journal of Intelligent Transportation Systems. He is Chair of the Transportation Research Board (TRB) Standing Committee on Connected and Automated Vehicle Systems and a member of the Board of Governors of the IEEE Intelligent Transportation Systems Society.

Agenda

The detailed agenda will be available later.

IEEE-T-ITS Special Issue Call for Papers

📌Align with this workshop, we are guest editing a Special Issue on AI-Empowered Automated Driving in Mixed Traffic: From Sensing, Perception to Planning and Control in IEEE Transactions on Intelligent Transportation Systems (IEEE T-ITS). For topics of interest and submission details, please check the Call for Papers.

Please prepare manuscripts according to the information for authors available at https://ieee-itss.org/pub/t-its/#toc_Submission_Information_Information_for_Authors, and submit via the IEEE Author Portal at: https://ieee.atyponrex.com/journal/t-its. When submitting, under “Article Type” in the portal, please select “Special Issue on Automated Driving in Mixed Traffic: From Sensing to Planning and Control”.

Resource Repository

The online resource repository for sharing relevant Datasets, Simulation Platforms, and Publications on Automated Mobility in Emerging Mixed Traffic can be accessed at https://qiqiqi.gitbook.io/mixed-traffic and https://github.com/IEEE-ITSS-OpenHub/Resource---Emerging-Mixed-Traffic-of-AV-and-HDV.

If you want to share relevant resources with the research community, please contact the workshop organizers.

Previous Editions 

At IEEE ITSC 2023 and 2024, the organizers hosted a previous edition of this workshop:  

https://www.itsc2024.mixedtraffic.org/ ,

https://sites.google.com/view/itsc2023-mixed-traffic.

Acknowledgment

The workshop is supported by the interdisciplinary research community and IEEE ITSS Technical Committees of Automated Mobility in Mixed Traffic and Cooperative & Connected Vehicles.