ITS Berkeley Collaborates with IPAM to Merge Traffic and Math

September 14, 2017

Several University of California Institute for Transportation Studies Berkeley faculty, staff and alumni will be working with the University of California Los Angeles’ Institute for Pure and Applied Mathematics (IPAM) merging traffic and math in a long program called New Directions in Mathematical Approaches for Traffic Flow Management.

The program consists of tutorials, a series of workshops and a culminating workshop at UCLA’s Lake Arrowhead Conference Center over four months in fall 2015.

Looking at the connection between math and how it’s used in the transportation field, faculty and staff from University of California, Berkeley, French Institute for Research in Computer Science and Automation, University of Texas at Austin, Technische Universiteit te Delft, French Institute of Science and Technology for Transport, Spatial Planning, Development and Networks, Northwestern University, University of Illinois at Urbana-Champaign and IBM wrote a proposal for the workshop and funding, partnering with one of the leading institutes in the world on applied mathematics IPAM.

Included in that team of organizers are ITS Berkeley Director Alexandre Bayen and alumni Chris Claudel (University of Texas at Austin) and Dan Work (University of Illinois at Urbana-Champaign). Other ITS Berkeley members involved throughout the series are ITS Berkeley faculty Roberto Horowitz, Alexander Skabardonis, Francesco Borrelli, and Pravin Varaiya; staff Gabriel Gomes and Alex Kurzhanskiy; and alumni Yafeng Ouyang (University of Illinois at Urbana-Champaign), Jorge Laval (Georgia Institute of Technology), Nikos Geroliminis (École Polytechnique Fédérale de Lausanne), Sebastien Blandin (IBM), and Ajith Muralidharan (Sensys Networks).

This series of workshops will look at how recent emergence of new technologies, such as sensor networks, smartphones, and new paradigms, such as crowdsourcing social networks, are transforming the way traffic management will be performed in the future. New advances have created the need for new modeling approaches (in particular to encompass the new data), new estimation, inference and filtering methods and are leading to the development of new paradigms for control. This revival of traffic engineering at the age of web 2.0 and social networks has created a significant amount of excitement in the mathematics, applied mathematics and engineering communities, to support these new approaches, and this program aims to capture these breakthroughs and bring together the world experts of these cross-disciplinary fields.

The program kicks off with Mathematical Approaches for Traffic Flow Management Tutorials from Sept. 9-12. It continues with Mathematical Foundations of Traffic from Sept. 28-Oct. 2. This workshop brings together traffic engineering and mathematics communities. The mathematics community has historically provided the engineering community with the proper ways to scientifically derive results used in practice, and the engineering community has provided the mathematics community with a variety of interesting problems to study. In the 21st century, the amount of work to be done on the mathematical side to provide a sound basis for the current work in engineering is considerable. It is growing due to new sources of data (such as smartphones) that have generated even more complicated problems.

The program continues with Traffic Estimation from Oct. 12-16. The workshop will investigate techniques that are commonly used for traffic estimation with partial differential equations, ranging from centralized and decentralized nonlinear extensions of Kalman filtering to particle filter; focus on statistical methods, in particular for the arterial networks where data is often sparse; optimization methods and games applied to networks of PDEs, with specific emphasis on traffic models; how to estimate mobility patterns on these networks, using massive datasets generated from call detail records and other positioning data; estimation problems such as tracking and localization, generated at the scale of small groups of pedestrians and crowds; and fundamental challenges and new approaches to maintain privacy of users who contribute the data to be used for estimation.

The next workshop is Traffic Control on Oct. 26-30. The workshop will explore the mathematical foundations of PDE control as the foundational theoretical work on which control of traffic systems rely. It will cover the fundamental approaches which are typically used for controlling PDEs (and which appear in the traffic control literature), in particular: Lypaunov stability, differential flatness, adjoint based optimization and more. Network control will also be covered, emphasizing coupling aspects, which appear during estimation, control or optimization on networks of PDEs (such as the transportation network). Routing, which is gaining attention due to the availability of smartphones that have routing capabilities and problems arising in future traffic due to the introduction of a number of VACS (vehicle automation and communication systems) round out the workshop.

Decision Support for Traffic from Nov.16-20 continues the program. The next decade will see numerous decision support tools emerge for traffic management, including sensing, communication, high performance and modeling capabilities. All over the world, transportation departments have started to investigate required steps to build tools and modeling capable of advising humans in charge of mobility optimization on a city scale and those experts who will present their experience. The workshop also will focus specifically on dynamic traffic assignment. The workshop will wrap up with experts in the field of games and incentivization, which is one of the future backbones of control and ties with decision support.

The program ends up with a Culminating Workshop Dec. 6-11, 2015.