Self-organization in Pedestrian and Traffic Systems and Logistics MOOC

•3 Modules, each consisting of several videos
•Estimated: 4 work weeks, 1h per week
•Self-paced, progress at your own speed
•No cost to enrol
•Subject: Computer Science, Traffic Systems, Social Science
•Level: Introductory
•Language: English
•Target groups: Students, citizen scientists, politicians, journalists, researchers of different fields
(urban planners, architects, computer scientists)
•Recommended Reading: Helbing, Dirk. Next Civilization: Digital Democracy and Socio-Ecological Finance-How to Avoid Dystopia and Upgrade Society by Digital Means. Springer
Nature, 2021

•understand the state of the art theoretical models applied in the field of traffic system research
•construct sustainable machine learning algorithms
•DIY urban simulation

•Prof. Dr. Dirk Helbing
•Javier Argota Sanchez-Vaquerizo (M.Arch., M.Sc.)
•Marcin Korecki (M.Sc.)

This lecture is an introduction to pedestrian and traffic systems with a focus on self-​organization. The lecture is divided into three parts. The first part deals with pedestrian systems and comprises three videos covering topics ranging from self-​organization in pedestrian crowds, spatio-​temporal pattern formation and crowd disasters

•external pageVideo 1call_made deals with emergent phenomena, namely interactions between pedestrians lead to certain patterns of collective motion, e.g. lane formation, bottlenecks, widenings and stripe formation. The social force model and more concretely the model-driven approach is introduced to reproduce and explain the previously seen emergent phenomena.


•external pageVideo 2call_made continues with a data-driven approach to the social force model. Simulation experiments on single pedestrians as well as on interactions between pedestrians are conducted to evaluate the social force model. Furthermore, video 2 addresses which kind of phenomena can be reproduced by the social force models. Insights from the social force model are discussed in connection with improving pedestrian facilities.


•external pageVideo 3call_made deals with the question of why crowd disasters happened, for example, during the Muslim pilgrimage in Mecca or the Love Parade in Duisburg. Insights from these two crowd disasters show that stop-and-go waves preceded the disaster. Furthermore, the notion of crowd pressure is introduced as an indicator for the likelihood of a crowd disaster occurring. Lastly, the video shows how these insights and the use of digital assistants can help to reduce risks.

The second part of this module deals with traffic systems and comprises three videos covering topics ranging from phantom traffic jams on freeways to adaptive cruise control as a possible solution. Somewhat similarly, traffic congestion in urban areas can be reduced by an adaptive traffic light control that promotes self-organized green waves.


•external pageVideo 4call_made deals with the formation of phantom traffic jams on freeways. It looks into complex congestion patterns, how they can be understood as a composition of elementary patterns and how to get rid of congestion. The Intelligent Driver Model (IDM) is introduced for traffic modelling and simulation. Then, unstable and metastable traffic states are discussed and how they may depend on the size of traffic disruptions.


•external pageVideo 5call_made dives into phase diagrams of congested traffic states to represent possible outcomes of (models of) freeway traffic. This gives insights into why traffic flow breaks down and how to design new adaptive cruise control systems based on real-time feedback and local interactions without the need of a traffic control center.


•external pageVideo 6call_made asks whether similar principles for dissolving traffic jams on freeways can be applied to reduce congestion in urban areas. Inspired by oscillatory pedestrian flows at bottlenecks, an adaptive traffic light control is proposed that results in self-organized green waves and non-periodic solutions. The revolutionary idea behind the proposed traffic light control is that the traffic flows control the traffic lights and not the other way round. Three different network approaches, including the proposed adaptive traffic light control, are compared. It is shown that a bottom-up self-regulation approach can outsmart centralized top-down control. The video ends with a case study of self-organized traffic light control in Dresden (Germany).

The third part of this module offers a wider perspective on the topic.


•external pageVideo 7call_made provides an outlook on urban planning, self-organization, and decentralization. Taking inspiration from the previously proposed self-organization approaches, the idea promoted in this last video is to turn our wasteful economy of today into a circular economy via real-time feedback and a multi-dimensional incentive system.

This module looks into the opportunities and challenges related to traffic light control in the age of artificial intelligence.


•external pageVideo 8call_made in this video we investigate, which problems need to be addressed in a smart city. Traffic light control is among the most critical ones.


•external pageVideo 9call_made in this video, we proceed by presenting and comparing an analytical and a machine learning approach to tackle the problem.


•external pageVideo 10call_made in this outlook video, we will will investigate how one can make machine learning more adaptable while still sustainable.


 

The goal of this module is to make an accessible urban simulation for experts as well as non-experts (e.g., technical people not used to simulations such as computer scientists, engineers, urban planners; other related professions like policymakers, journalists, and government officials; and citizens in general). This module consists of a guided tutorial supported by a Jupyter notebook, where the needed code is shared to complete a whole small simulation from pulling data, filtering, deploying, visualizing and testing what-if/alternative scenarios that can be of interest to the prospective audience. Finally, using that framework, some examples for testing alternatives (based on A/B testing) will be discussed to introduce potential advantages of extending these tools for informing parties involved in city co-creation processes and how they can improve the quality and engagement in participatory processes.


•external pageVideo 11call_made introduces the use of computer simulations as an enabler and facilitator of informed participatory processes. Then, it guides the audience how to set up an initial basic simulation based on the repository shared with this block course.


•external pageVideo 12call_made expands on basic computer simulations to test different alternatives, whose data and configuration files are available in the shared repository as well, including visualizations of simulation results. The comparison of these what-if scenarios is done in a way that allows one to explore possible future situations. Finally, this is used to exemplify how computational tools can help to enable a more comprehensive, interactive, and transparent way of participatory city co-creation.