Modelling traffic disruptions using machine learning and simulation modelling

Project Topic
Traffic incidents are among the primary concerns of all transport authorities around the world due to their significant impact in terms of traffic congestion and delay, air and noise pollution, and management cost. This PhD project aims to study and model the impact of traffic disruptions and various response plan evaluations by combining multidisciplinary research from transport modelling and data science. The results are expected to be applied for both road traffic and public transport modelling in real-life setups.

The first part of the PhD will focus on deploying several machine learning or deep learning models combining classification and regression approaches for incident duration prediction and early traffic anomaly detection (see our recent work [1], [2], [3]). While recurrent traffic modelling can provide good insights on traffic patterns and easy anomaly detection, incidents are stochastic events which have unique features changing dynamically in time. The second part of this project will aim at developing and modelling the impact of accidents by using traffic simulation models adapted to the level of study (micro/meso simulation modelling or agent-based). Lastly, this approach is expected to be linked as well with various optimisation techniques for evaluating multiple response plan scenarios and provide recommended actions for dealing with future similar disruptions (see our work on integrated data and traffic modelling published in [4]).

About the PhD
The PhD student will be located within the newly formed Future Mobility Lab at UTS cofounded by Dr. Simona Mihaita, under a new Data Science Institute led by Distinguished Prof. Fang Chen. The institute counts around 30 staff members with research interests spanning across asset management, transportation, behavioural data science and human dynamics. The Data Science Institute has both strong ties with industry, as well as world-class research, providing the ideal environment for solving real-world problems, in close proximity to both academia and industry. The candidate: Interested candidates must have solid background knowledge in transport modelling, machine learning, data science and strong programming capabilities in Python/R. Experience with handling large and complex transport data sets and transport modelling at micro and meso levels are a big plus. We are looking for a candidate with a master by research qualification and demonstrated research capabilities (preferably through publications). Candidates with publications in major conferences/journals will be prioritised.

The position will be open until the ideal candidate is identified. In order to complete your application, please send us:
your CV
- grades transcripts from undergrad and Masters
- your research proposal ideas on the topic (max. 2 pages)
- Masters thesis manuscript (if applicable) or any other research thesis;
- a cover letter (max 1 page), outlining how your profile fits the PhD position;
- 3 referees (academic/industrial supervisors, co-authors): name, position and email;
- (if relevant) one of your publications.

The selected PhD student will work under the supervision of Dr. Mihaita and interact closely with academics under the ARC Linkage Project LP180100114 recently awarded through the Australia and Singapore Strategic Collaboration. Partners: Swinburne University of Technology, Data61

CSIRO, National University of Singapore and National Technology University from Singapore. Regular meet-ups and workshops will be organised for presenting new findings and learn from new techniques applied both in Australia and Singapore.

References:
1. Mihaita, A.S., Liu, Z., Cai, C., Rizoiu, M.A “Arterial incident duration prediction using a bi-level framework of extreme gradient-tree boosting”, ITS World Congress 2019, Singapore, 21-25 Oct 2019, Preprint: https://bit.ly/2FjBx4m
2. Mao, T., Mihaita, A.S., Cai, C., Traffic Signal Control Optimisation under Severe Incident Conditions using Genetic Algorithm, ITSWC 2019, Singapore, 21-25 Oct 2019, Preprint: https://bit.ly/2ITBCwF
3. Shaffiei, S. Mihaita, A.S., Cai, C., Demand Estimation and Prediction for Short-term Traffic Forecasting in Existence of Non-recurrent Incidents, ITSWC 2019, Singapore, 21-25 Oct 2019,Preprint; https://bit.ly/2ZtSOzf
4. Wen Tao, Mihaita A.S., Nguyen Hoang, Cai Chen, Integrated Incident decision support using traffic simulation and data-driven models. TRB 97th Annual Meeting, Washington D.C., 2018, https://bit.ly/2IMSe98