苏黎世联邦理工

Deep Generative Behaviour Models for Travel Demand Forcasting

项目介绍

The Singapore ETH Centre Future Cities Lab Global, in collaboration with the National University of Singapore and the Graduate School of Advanced Science and Engineering at Hiroshima University, has an opening for a PhD Researcher in agent-based, activity-based transport demand modelling.

Project background

In recent times, cities have experienced an unprecedented period of flux, whether triggered by disruptive technology, political upheavals, emergent social trends or the ongoing global pandemic. These growing uncertainties have prompted cities to rethink land use and transportation infrastructure provision planned for the future. The ETH and NUS, in collaboration with a team at Hiroshima University, have a new project, Adaptive Mobility, Infrastructure and Land use, “AMIL”, within the Future Cities Lab – Global Research Programme (FCL-G) that aims to integrate mobility, land use and infrastructure into a resilient, adaptive system that responds to changing needs, through a three-pronged approach.

First, social network surveys in Zurich and Singapore will monitor and examine the ‘drivers of change’ by assessing changes in social and economic activity, further informed by the ongoing experiments at Hiroshima University in the joint activity decision-making process.

Second, novel transport modelling and simulation techniques will be developed to handle the unique challenges posed by the ‘city in flux’, with dynamically adaptive land use and demand-responsive transportation systems.

Third, exploratory modelling, decision-making under uncertainty methods (e.g., real-options) and optimisation methods will be employed to formulate adaptive plans that can respond to short-term and long-term change.

The project will bring tools, insights, methods, and procedures that can provide decision support for planning across relevant scales of time and space. Simulation models and software tools will be developed in consultation with an expert panel of practitioners and policymakers, to enable adaptive planning, improve resilience to shocks such as COVID-19, and reduce misalignment of provision and emerging demand in the long-term. These learnings will be applied to a study of urban development across intercity, city-wide and local scales in both Alpine and Asian contexts, with Zürich, Singapore and Higashi-Hiroshima as the main hubs under study. The project will be guided by regular stakeholder workshops in both contexts. It will produce a suite of open-source software tools and planning recommendations developed in collaboration with an inter-agency working group for adaptive mobility, land use, and infrastructure.

Job description

You will work on designing innovative experiments to understand social network effects in individual-level daily activity choices, advancing travel behaviour models, and developing new deep learning techniques to generate synthetic population and its activity patterns while accounting for social network effects. You will assist in incorporating these models into an urban scale agent-based travel demand forecasting framework, MATSim. The detailed tasks would be as follows:

  • Design experiments to understand how people coordinate activity time, location, and travel mode choice with other household members and friends. Create a web-based platform to conduct such experiments in Singapore. Develop a new set of choice models which can account for social network effects in estimating activity-based travel preferences.
  • Develop deep generative models to combine the collected data with the large-scale household travel survey data of Singapore to generate a synthetic population of Singapore with the social network effect.
  • Integrate social-network-based synthetic population and choice models with other data sources (e.g., cellular phone traces) to generate activity plans at an urban scale. Advance state-of-the-art methods to predict the secondary activity locations.   

You will be supervised by the principal investigator at the National University of Singapore, Assist. Prof. Prateek Bansal, will lead the social network survey and collaborate on the demand generation and simulation packages. You are also expected to work closely with the co-investigator at Hiroshima University, Adj. Prof. Pieter J. Fourie, and his team, lead MATSim development for the project. Other colleagues in the team at FCL-G will focus on the estimation of meta-models from ensemble MATSim runs to facilitate the development of a long-term infrastructure planning framework to deal with multi-level risks and uncertainty. Additionally, You are expected to participate in regular exchanges and coordination with Prof. Kay Axhausen and his research team in Zürich, who will be working on similar topics.

Your profile

  • A bachelor’s/master’s degree in engineering, statistics, economics or a related field.
  • A strong background in probability theory, model-based machine learning, statistics, big-data processing pipelines, high-dimensional data analysis, and discrete choice models.
  • Proven experience and strong programming skills in Python and R.
  • Proficiency in written and spoken English.

项目概览

wave-1-bottom
访问项目链接 招生网站
欧陆, 瑞士 所在地点
带薪项目 项目类别
截止日期 2022-09-04
苏黎世联邦理工

院校简介

苏黎世联邦理工是国际研究型大学联盟、全球大学高研院联盟、IDEA联盟成员,是闻名全球的世界顶尖研究型大学,连续多年位居欧洲大陆高校翘首。
查看院校介绍

联系方式

电话: +41 44 632 11 11

相关项目推荐

KD博士实时收录全球顶尖院校的博士项目,总有一个项目等着你!