Symposium on activity-based modeling

From September 12th to 14th, 2022, a symposium on activity-based models will take place the Kloster Seeon. This symposium is only by invitation. The symposium results and presentations will be published in this website after the event. 

There are about ten operational activity-based models (ABM) today. The understanding of the limitations of trip-based models, availability of new data sources and increased computing power have encouraged more researchers and agencies to move towards ABMs. However, ABMs are not as widespread as their potential would suggest. ABMs require extensive calibration, tend to be driven by constants, lead to comparatively long runtimes and suffer from stochastic differences between runs.

The advent of new transport modes and digitalization, such as autonomous vehicles, electric vehicles, ride-hailing, ride sharing, telework or e-commerce affect daily activities, make ABMs more relevant than ever before. The COVID-19 pandemic implies that transport models also need to understand social networks to better understand who meets whom on a regular basis to model disease transmission. It is time to rethink what the new generation of ABM needs to accomplish, and how it can be realized.


Syposium objectives

During this workshop, the participants will discuss the next direction of activity-based models (ABM) to try answer the following questions: 

  • How can ABM applied to multiple years, in connection with land use models? How can they re-use activities and plans of previous years? Will this improve the model fidelity and the runtime? 
  • How different types and frequencies of activities (and the ABM used to represent them) corresponf to the short-distance and the long-distance travel segments? 
  • Are ABM capable to represent social networks? Does this influence their results? 
  • Can we model health impacts based from an accurate simulation of activities? 
  • Can ABM model the trips and activities made by tourists and visitors? 
  • What do we need to improve the computational efficiency of large-scale ABM simulations?