Video Player

Traffic Paradoxes and Route Guidance: Effective Ways of Reducing Congestion Effects?

Andreas S. Schulz: December 1, 2009; Running Time: 58:18

About the Lecture

About the Lecture

It is well know that we cannot engineer our way out of traffic congestion by building new roads. In fact, expanding the road network may paradoxically attract new traffic, and increase gridlock. Andreas Schulz provides a mathematical explanation for this conundrum. Using Nash equilibria and related game-theoretic concepts he explores two issues, namely: “how much fuel and time can we save if we route traffic optimally, and secondly, can we save fuel and time by actually closing streets or rearranging vehicle flow on our existing road network?” The answers to these questions have significant value. It is calculated by the Texas Transportation Institute (TTI) the cost of congestion, in fuel and time losses, is $87 billion annually (in 2007 dollars). Schulz uses the TTI estimate as a launching point, to ask how much we could save if we routed more optimally.

The optimization is based on a complex set of algorithms with Wardrop’s Principle as a theoretical background, and total travel time as the variable. Wardrops principle says that the journey times on all the routes actually used are equal and less than those that would be experienced by a single vehicle on an unused route. From this user optimum/equilibrium, Schulz branches to a key concept called the “price of anarchy”. Applied to traffic, it is a ratio of the journey time of the individual transport user (represented in the numerator) to the value of a system optimization (in the denominator). The so-called “price of anarchy’ relationship, numerically expresses what is lost in terms of travel efficiency when each driver acts on their selfish interests (autonomously) instead of using the optimized network.

Schulz uses scenarios about traffic delay and travel times and the mathematical proof shows that the price of anarchy can be measured and valued. A simple graphic establishes the relationship between the travel time function for individual actors vis-à-vis those of the collective. An adjustment is made for free-flow versus peak traffic to reflect a steep, non-linear incline between the number of peak hour vehicles and road congestion. Schulz finds that his two initial questions are theoretically linked-- knowing the first optimization helps reach the second one, namely the outcome from closing roads. He notes that the equations can enumerate savings from closing roads that handle different levels of traffic volumes. But he adds, these equations cannot tell the researcher “which roads to close ”.

While solving for full optimization, Schulz also proposes there can be a different path, namely to provide a constraint system optimization. Here, individual drivers are assigned to paths that are in his words “ not too long”. Instead, there is a ‘fairness’ applied to the path allocation, so that most drivers are generated a travel route within normal driving ranges or boundaries. Using this constraint systems approach, the optimizations reveal that 75% of the road users would experience less travel time in lieu of their individual route choice.

The take-away, says Schulz, is that we can actually route traffic much better than we currently do and adding additional roadways to a system will not necessarily make it better or alleviate traffic. The simulations made for Boston’s Big Dig bear this out, and he proposes that there are many other practical applications, particularly using the constraint systems. Current navigation systems, programmed at a system wide level, show great promise for shortening our travel time, if not occasionally lengthening our travel path.

Lecture Details

Location: 32-124

“Can we save fuel and time by actually closing streets or rearranging vehicle flow on our existing road network?”

Andreas Schulz

About the Speaker

Andreas S. Schulz

Patrick J. McGovern (1959) Professor of Management
Professor of Mathematics of Operations Research

Andreas S. Schulz is the Patrick J. McGovern (1959) Professor of Management and Professor of Mathematics of Operations Research at MIT. He is also a faculty member of MIT's Operations Research Center, and he has held visiting research professorships at Maastricht University and ETH Zurich.

He received a Ph.D. in Mathematics from the Technische Universität Berlin in 1996. His research interests include algorithmic game theory, approximation algorithms, combinatorial optimization, computational complexity, integer programming, network flows, polyhedral combinatorics, and scheduling. In 2000, the German Academy of Sciences Leopoldina and the Berlin-Brandenburg Academy of Sciences and Humanities named him one of 20 founding members of "Die Junge Akademie." Other honors include the Best Paper Award of the Transportation Science & Logistics Society of INFORMS, the Glover-Klingman Prize, the Carl-Ramsauer-Prize, and awards for excellence in teaching. He has been on the editorial boards of several journals, including ACM Transactions on Algorithms, Discrete Optimization, INFORMS Journal on Computing, and Operations Research.

Andreas S. Schulz

Schulz home page

About the Host

About the Host

Transportation@MIT

Transportation@MIT is a coordinated effort to address one of civilization’s most pressing challenges: the environmental impact of the world’s ever-increasing demand for transportation.

Building on MIT’s rich tradition of engineering research and interdisciplinary collaboration, the initiative knits together the wide-ranging, robust research already under way at the Institute and creates new opportunities for education and innovation.