With urbanization and centralized economic activities, recent years have witnessed steady population growth in major cities around the globe. For example, the population in the city of Zurich has increased by 16% between 2000 and 2020, which certainly leads to an ever-growing traffic demand, potentially more traffic accidents, and increasingly congested networks.
The concern is further exacerbated by our recent research findings, revealing that road transportation networks exhibit the property known as fragility. Such a fragile nature indicates that, with a linearly growing magnitude of disruptions, the collective loss in the system performance grows exponentially. For example, the initial 10% increase in demand can lead to a 10% worsening of congestion, but each subsequent 10% increase will cause progressively greater impacts, such as 20%, 30%, 40%, and so on.
Quantifying Network Fragility
Based on established macroscopic traffic models, we further investigate the impact of certain traffic-related variables with physical meanings on such fragility of road transportation networks. To achieve this, a quantitative indicator derived from distribution skewness has been developed. The results suggest several policy-relevant implications: imposing stricter speed regulations or introducing autonomous vehicles can reduce network fragility by lowering free-flow speed or increasing backward wave speed, respectively. Additional findings also indicate that real-world stochasticity has a reinforcing effect on the fragile characteristics of road transportation networks.
Designing Antifragile Solutions
The solution to fragile systems is antifragility, which was first proposed by Nassim Nicholas Taleb in his bestseller Antifragile: Things That Gain from Disorder. In our ongoing work, a reinforcement learning-based traffic control algorithm has been proposed. Under a numerical simulation environment, the algorithm leverages historical disruptions of lower magnitude to achieve better performance when faced with increasingly severe disruptions, simulating either demand disruptions such as large-scale population gatherings or supply disruptions such as lane closures due to natural hazards.
Prospect
Under the EU Horizon Project AntifragiCity, the proposed antifragile algorithm will be further validated with real-world networks and data across different European demonstration cities, including Bratislava, Larissa, and Odessa. Such a philosophy of the fragile nature and antifragile solutions can be expanded into other transportation modes and potentially other disciplines of control systems and engineering.
Linghang Sun is a doctoral student at the Institute for Transport Planning and Systems at ETH Zürich. His research focuses on developing the concept of system antifragility in the context of transportation. In particular, his research interests include traffic flow theory, traffic control, reinforcement learning, and transportation safety. He is responsible for the ongoing AntifragiCity Horizon Europe project.
Find out more
New Project to Foster Antifragility for Urban Traffic Control
» Newsletter-articles
Antifragile Urban Mobility: Traffic Control Beyond Resilience
» Newsletter-articles
