ITSC2024
Keynote Speakers
Agenda
Agenda at a glance:
[Local time 8:50~12:30 (MDT, UTC-6), September 24, 2024 | Room: Salon 5]
Moderator: Zhe Fu and Yongqi Dong
➢8:50 – 8:55: [5min] Opening: Workshop aim & agenda
➢8:55 – 9:10: [15min] Ice breaker: Use https://vevox.app ID: 111185169 (https://vevox.app/#/m/111185169) to know participants.
➢ 9:10 – 10:30: [50 min] Keynote presentation - session 1
· 9:10 – 9:35 [25min] Dr. Maria Laura Delle Monache (UC Berkeley), Control Strategies for Mixed Autonomy Traffic Systems
Abstract:
In this talk, Maria Laura will focus on techniques to manage and control mixed autonomy traffic flow. Maria Laura will show control strategies for traffic systems with the aid of fleets of connected and automated vehicles immersed in human-driven traffic flow. We will prove analytically and numerically how the proposed control theory can improve traffic performance, and Maria Laura will present an open field test involving 100 connected and automated vehicles (CAVs).
Speaker's Bio:
Maria Laura Delle Monache is an assistant professor in the Department of Civil and Environmental Engineering at the University of California, Berkeley. Prior to joining the faculty at UC Berkeley, she was a research scientist at Inria in Grenoble, France (2016-2021) and a Postdoctoral fellow at Rutgers University - Camden in USA (2014-2016). She received the Ph.D. degree in applied mathematics from the University of Nice-Sophia Antipolis, France in 2014. She is a member of the IEEE CSS Technical Committee on Smart Cities and of the Standing Committee on Traffic Flow Theory and Characteristics of the Transportation Research Board (NASEM). She is the recipient of the 2023 IEEE TCCPS Mid-career award, the 2023 IEEE ITSS Young Researcher/Engineer award, and the 2024 UC ITS Faculty of the year award. Dr. Delle Monache’s research lies at the intersection of transportation engineering, mathematics, and control theory.
· 9:35 – 10:00 [25min] Dr. Cathy Wu (MIT), Model-Based Transfer Learning for Mixed Autonomy Traffic
Abstract:
Critical to designing future mixed autonomy traffic systems is solving many hard optimization and control problems. As such, researchers increasingly look towards data-driven methods such as deep reinforcement learning (RL). However, one issue that limits RL's practical application is its brittleness, sometimes failing to train in the presence of small changes in the environment. We address this challenge by leveraging the empirical observation that applying an already trained model to a related task often works remarkably well, also called zero-shot transfer. We take this practical trick one step further to consider how to systematically select good tasks to train, maximizing overall performance across a range of tasks. Given the high cost of training, it is critical to choose a small set of training tasks. The key idea behind our approach is to explicitly model the performance loss (generalization gap) incurred by transferring a trained model. We hence introduce Model-Based Transfer Learning (MBTL) for solving contextual RL problems, which emits several effective algorithms. Despite the conceptual simplicity, the experimental results suggest that MBTL algorithms outperform strong baselines.
Speaker's Bio:
Cathy Wu is an 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 aims to leverage machine learning to solve hard optimization problems for next-generation mobility systems. She is broadly interested in leveraging modern computing and AI to advance decision making. 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 AC/AE for ICML, NeurIPS, and ICRA. She is also helping spearhead efforts towards reproducible research in transportation.
➢ 10:00 – 10:30 [30 min] Coffee Break & Speed Dating
· During the break, the audiences are encouraged to talk to people with different colors of cards (indicating different research domains) assigned at the beginning
➢ 10:30 – 11:20: [50 min] Keynote presentation - session 2
· 10:30 – 10:55 [25min] Dr. Daniel Work (Vanderbilt University), Variable Speed Limit Control of Freeways with Automated Vehicles
Speaker's Bio:
Dan Work is a Chancellor Faculty Fellow and professor in civil and environmental engineering, computer science, and the Institute for Software Integrated Systems at Vanderbilt University. He has held research appointments at the University of Illinois at Urbana-Champaign (2010-17), Institute for Pure and Applied Mathematics (2015, 2020), Microsoft Research Redmond (2009), and Nokia Research Center Palo Alto (2007-09). Dr. Work received a 2018 Gilbreth Lectureship from the National Academy of Engineering and a 2014 CAREER Award from the National Science Foundation. He earned a BS from Ohio State in 2006, and an MS (2007) and Ph.D. (2010) from UC Berkeley, all in civil and environmental engineering.
Dr. Work pioneered methods for monitoring and controlling road traffic using vehicles, rather than fixed infrastructure, to sense and control road congestion. In 2015 he and his collaborators were the first to experimentally demonstrate that "phantom" traffic jams, which seemingly occur without an obvious cause but are due to human driving behavior, can be eliminated via control of a small fraction of automated vehicles in the flow. Work is a recognized transportation expert whose work has appeared in media outlets including ABC's Good Morning America, Reuters, Wired, and MIT Technology Review.
· 10:55 – 11:20 [25min] Dr. Lina Kattan (University of Calgary), A Data-Driven Approach for Urban Road Network Resilience Assessment: Integrating Spatiotemporal Analysis with the Resilience Triangle Concept
Abstract:
Lina will present a new data-driven approach to investigate the spatiotemporal impact of daily non-recurring disruptions and measure the recovery time, vulnerability and resilience of links in urban road networks. Multi-year observed travel time and incident data are utilized to capture the impact of incidents on the network, considering the inherent interconnectedness of the network components. The study develops a statistical method to estimate event occurrence, network restoration, and incident recovery times and formulates a new time-dependant efficiency function borrowed from complex network theory. Using historical travel time and incident data from the City of Calgary, Canada, this study applies the developed methodology to measure network’s response to disruptions and assess the resulting link resilience loss. The developed efficiency function successfully estimates the spatiotemporal change in network performance and reconstruct the resilience triangle, which is used in turn to measure the vulnerability and resilience loss for each impacted link. The results of the analysis indicate that the actual occurrence time of the incident is often earlier than the time it is reported. In addition, the results indicate that low resilience links tend to be geographically clustered, often near high-demand generation/attraction regions with low network redundancy.
Speaker's Bio:
Lina Kattan is a Professor of Transportation Engineering at the Schulich School of Engineering, University of Calgary. She holds the prestigious Canada Research Chair (CRC) Tier I in Integrative Transportation Systems through Automation and Connectivity and the Urban Alliance Chair in Transportation Systems Optimization. Additionally, she serves as the Director of the NSERC CREATE program in Integrated Infrastructure for Sustainable Cities (IISC), a pan-Canadian collaborative research initiative dedicated to developing systematic solutions for the multifaceted needs of future cities.
Lina's research encompasses traffic flow theory and control, transportation network modelling and analysis, equity and fairness in transportation, emerging vehicular technologies, and public transit operations.
➢ 11:20 – 12:20 [~60 min] Panel discussion, joint Q&A
· Panel Discussion: Maria Laura Delle Monache, Cathy Wu, Lina Kattan [~45 min]
- Emerging critical issues in mixed traffic
- Data-driven research and reproducibility
- Policy and recommendations
· Q&A [~15 min]
[This session is a joint effort with the tutorial “Reproducibility in Transportation Research: A Hands-on Tutorial”.]
➢ 12:20 – 12:25 [5 min] Wrap up & relevant resource sharing (Github of datasets, simulation tools, key publications)
and introduction of Automated Mobility in Mixed Traffic TC Committee
➢ 12:25 – 12:30 [5 min] Closing the workshop, future events, and casual networking
➢ Afternoon joint tutorial session: “Reproducibility in Transportation Research: A Hands-on Tutorial”
This session is dedicated to exploring the role of data management in fostering reproducibility and credibility in transportation research, with a focus and examples on mixed traffic data.
· Presenting the Findability, Accessibility, Interoperability, and Reusability (FAIR) principles
· Empirical findings
· Limitations of current datasets, data format standards, and privacy-preserving techniques
· Research needs, challenges, and future research direction
· Prediction for the development and deployment of Automated Mobility
