Latest updates
  • Apr 25, 2020: web site of the challenge opens, the task is revealed,

Understanding individual and crowd dynamics in urban environments are critical for numerous applications, such as urban planning, traffic forecasting, and location-based services. Benefiting from the explosive of data collection techniques, human mobility can be observed, analyzed and modeled from GPS, CDRs, IC card Records and et al. However, data collection is rather demanding and difficult for some developing countries and local areas, and the use of such kinds of data has complicated issues of legal rights to privacy.

On the contrary, modeling and simulating human mobility at both the individual and population levels simultaneously is a promising approach to overcome this issue. Recently, researchers from different domains have proposed lots of different approaches to detect, classify and predict human mobility. It is essential to develop a method which fuse these various aspects of human mobility and reconstruct movement behavior which is available to be open to public.

This task is to provide a large amount human mobility dataset at Tokyo, Japan, and challenge participants to create synthetic people mass intracity movement in atypical day in Osaka, another metropolitan area in Japan. Unlike other popular open mobility dataset, the “People Flow Dataset” consist of large population (about 5% sampling rate over population) which enable participants to freely develop and evaluate the model and output. The generated synthetic datasets with a considerable accuracy will be used as a replacement to visualize, analyze and predict the citywide level individual crowd dynamics.

The Task

Participants will challenge to propose an algorithm that can model within-day human mobility at citywide level from training dataset at Tokyo, Japan. Using this algorithm, participants are required to construct a synthetic mobility dataset that report each individual's (given its initial state) to simulate people mass daily movements at Osaka, Japan.

After the competition phase is completed, a link for the submission of the accompanying academic paper will be provided to the top 10 participants as ranked by the public/private leaderboard weighting.

External open materials (i.e. National Census, Time Use Survey, Infrastructure Information or any other open datasets) are also allowed to be introduced to this work.

  1. Sekimoto, Y., Shibasaki, R., Kanasugi, H., Usui, T., & Shimazaki, Y. (2011). Pflow: Reconstructing people flow recycling large-scale social survey data. IEEE Pervasive Computing, 10(4), 27-35.
  2. Kashiyama, T., Pang, Y., & Sekimoto, Y. (2017). Open PFLOW: Creation and evaluation of an open dataset for typical people mass movement in urban areas. Transportation research part C: emerging technologies, 85, 249-267.