The CASE (Connected, Automonous, Shared, and Electric) mobility will greatly change transport servises. One illustrative example is self-driving vehicles with multiple functions such as ride-sharing, e-coomerce, and logistics, which would enrich our daily lives. Various personalized services would be offered based on the enormous data from vehicles mobile phones, etc. At the same time, such big and real-time data would also change transport management systems drastically together with the rapid development of relevant methodologies.

One of the key common ingredients for a better mobility service and its management systems is the short-term prediction of transport conditions: an accurate short-term prediction of OD demand and travel time would be needed for a better ride-sharing service, while a better short-term prediction of traffic states using real-timedata would significantly improve dynamic traffic control and management systems. One of the emerging and promising approaches for a better short-term prediction is a machine learning approach. Appliactions of machine learning techniaues in the field of transportation have been increasing rapidly in the last couple of years. These studies have empirically shown higher prediction accuracy compared to traditional methods, opening up further possibiilties of providing new transport services as well as data-driven traffic control and management.

This seminar aims to identify unique challenges in the application of machine learning techniques to the short-term prediction, explore further possibilities ot applying deep learning techniques to transport issues, and identify potential bottlenecks in utilizing them in practice. Following a special lecture of the use of tree search and deep neural networks by Dr. Yoshizoe, two keynote lectures will be delivered by Dr. Chris van Hinsbergen and Dr. Adam Pel on the state of the art for short-term traffic prediction in Netherlands. We will then have presentation from researchers and practitioners on their ongoing works and discuss the possible future research directions and practical applications.

Date and time: 10:00-17:30 on September 20, 2019

Venue: Meeting Room 3, Suekawa Memorial Hall, Ritsumeikan University, Kyoto

9 Kinugasa Himurocho, Kita-ku, Kyoto, 603-8484

(Map: https://goo.gl/maps/N2GrysfoJnb9YHaX9)

Capacity: 40 persons

Registration: Please send your name and affiliation to Makoto Chikaraishi

(chikaraishim@hiroshima-u.ac.jp)

Note: The application will be closed as soon as the number of applicants reaches the capacity.

Program

Project Introduction and Special Lecture

     Organizer: Yasuhiro Shiomi (Ritsumeikan Unievrsity)

10:00-10:15: Introduction of research project

“Short-term travel demand prediction and comprehensive transport demand management”

by Makoto Chikaraishi (Hiroshima University)

10:15-10:30: A brief overview of the application of machine learning models in the field of transportation

by Varun Varghese (Hiroshima University)

10:30-11:30: Special Lecture: Solving Problems Using Tree Search and Deep Neural Networks

by Kazuki Yoshizoe (Leader of Search and Parallel Computing Unit, RIKEN Center for Advanced  Intelligence Project)

11:30-13:00: Lunch break

Keynote Lectures

Organizer: Makoto Chiakaraishi (Hiroshima University)

13:00-14:00: Keynote lecture 1: Traffic Theory & Decision Forests for prediction of local traffic patterns

by Adam Pel (Associate professor, Delft University of Technology)

14:00-15:00: Keynote lecture 2: The Neural Cell Transmission Model

by Chiris van Hinsbergen (Co-Founder & Developer, Fileradar)

15:00-15:20: Coffee break

State-of-the-Art Research and Practice

     Organizer: Varun Varghese (Hiroshima University)

15:20-15:50: Traffic Congestion Control by Vehicle Trajectory Estimation

by Masaaki Ishihara (Hanshin Expressway Company Limited)

15:50-16:20: Short-Term Traffic State Prediction Using the LSTM Framework: A Case Study in Kamakura City

by Daisuke Fukuda (Tokyo Institute of Technology)

16:20-16:50: Toyota’s activities in MaaS

by Takahiro Shiga (Toyota Motor corporation)

16:50-17:20: Driver’s Behavior in Ride-hailing Service

by Junji Urata (The University of Tokyo)

17:20-17:30: Closing