ADAIT team in D61+Live Event
ADAIT team hold a booth in the D61+Live event at Brisbane Convention and Exhibition Center.
Our postdoc researchers Tuo Mao, Sajjad Shafiei, and software engineer Adrian Plani showcase some of the latest outcomes of ADAIT's innovations.
- Machine learning to predict 30 minutes of future traffic conditions
- An end-to-end multi-modal transport journeys platform
- A large-scale simulation platform covering the whole of Sydney
Huge thanks to all of the attendees of the D61Live event!
Data61 at World Transport Convention 2018, Beijing
by Dr Chen Cai
It was an honor be invited for a presentation in the International Forum in Big Data Leading the Innovation and Development in Transportation, a parallel session of the World Transport Convention 2018, Beijing.
Jointly organised by the China Academy of Transportation Science, a research agency for the Ministry of Transport, the Highway & Transportation Society and Beijing Join-Cheer Software Co. Ltd, the Forum focused on the development of transport data policies, platforms and their user stories. A total of 14 invited speakers plus a panel discussion filled a one-and-half day agenda. The speakers covered topics in transport data policies, standardization, security, emerging technologies from AI and Blockchain, and a range of applications in civil aviation, railway, logistics, planning and mobility services.
On behalf of Data61|CSIRO, I presented “On Demand Public Transport & Predictive Analytics”, a work that uses an AI engine to predict short-term travel demand for the Keoride services in Sydney’s Northern Beaches. Keoride is part of the ongoing TfNSW On Demand Public Transport trials that pilot a range of innovative operation models for mobility-as-a-service. A news article in Chinese on my presentation can be found here: http://www.chinahighway.com/news/2018/1176816.php
The development on the AI engine is a top priority for Advanced Data Analytics in Transport (ADAIT), a research group at Data61. ADAIT leverages its unique access to vast, granular, and real-time transport datasets to train and select machine learning algorithms that form the basis for the AI engine. ADAIT sees AI as the core technology to advance progresses in autonomous vehicles, intelligent mobility, autonomous control, and transport safety. The aggregated benefits of adopting AI in transport will lead to improved economic productivity and better urban liveability.
IoT Festival 2018
by Dr Simona
This week I had the pleasure to be a speaker in the "Connected and Autonomous vehicles" panel in the IoT Festival 2018 in Melbourne, together with Neil Wong from the National Transport Commission , Denise Christie from Intelematics and Sue Wiblin from Keolis Downer.
Various questions have been raised by the public especially as Autonomous Vehicles are a certitude now and will soon become a reality impacting the way we move through the city, the way we work and have access to goods or entertainment. I thought this is a good opportunity to wrap up some of the thoughts and ideas that have been exchanged during the event.
Some of the current struggles impeding an early adoption of CV/AVs are the following:
- lack of advanced road infrastructure to support a fast and efficient V2V (vehicle-to-vehicle) and V2I (vehicle-to-infrastructure) communication. The ability for new autonomous cars to communicate between them but also with current traffic lights or road side units is extremely important to re-evaluate their position, send and receive early collision alarms which would help prevent traffic incidents and inform traffic management centers of ongoing traffic anomalies. Dedicated lanes for CV/AVs could be a first step to ease travelling especially in a mixed traffic environment which will follow a slow integration curve of all new transportation modes.
- lack of standard adoption, insurance policies, safety regulations in various countries around the world. Australia is making its first steps towards early testing and adoption of CVs/AVs but current city infrastructures and long distances between large urban centers make it challenging to reduce the private car ownership in the future. Travel time and costs are two of the most important factors that citizens value when travelling on a daily basis. The new CV/AV mobility solution should bring either a significant reduction in travel time or smaller travel costs when compared to existent travel modes, either private or shared.
- lack of design thinking of the CV/AV adoption under the overview of a whole eco-system which should include various entities responsible with: production/ testing/ compliance checking/ software upgrade/ data collection and transmission/vehicle maintenance/placement of recharging hubs/repairing and replacement of damaged vehicles/ real-time traffic congestion awareness and updated urban maps/ road accident evaluation bodies to investigate the cause of potential accidents involving CAVs/etc.
- lack of research studies which would evaluate various aspects of the CV/AV adoption such as: a) smart data analytics for understanding how the travel demand will evolve in various regional urban areas affected by an on-demand and possible shared-autonomous transport system, b) travel behavior under recurrent and non-recurrent traffic conditions when multiple transport modes are available for use, c) optimal dynamic re-routing in case of incidents, d) impact of various pricing schemes on multi-modal choices, e) impact of replacing one transport mode with another, especially if this mode will be autonomous and/or shared, f) smart schemes for traffic control which would balance the supply and demand efficiently in the city, especially during peak hours.
- lack of regulatory bodies for dealing with data privacy, security breaches and safety compliance. These services would need to be created for normal use of CAVs but also for creating efficient response plans under emergency situations where special intervention would be needed. Government bodies would need to have access to all the data generated by CAVs in order to make better traffic management decisions and evaluate the impact of future planning operations in the city.
Some of the benefits of adopting CAVs in the future will highly be impacted by the scheme of early adoption and purchase of such vehicles. If they will be adopted in a car-shared mentality, some of the potential benefits are the following:
- faster travel times in central urban areas due to more parking slots not being used as all cars would be continuously circulating on the streets,
- more personalized services as some companies will choose to deliver door-to-door goods,
- more access to urban facilities for various kinds of travelers, especial for people living in remote areas who require access to health services, education, etc.
- a potential decrease in personal car ownership if the service proves to be fast, efficient, safe and cheaper (or similar in price) than existing modes. While the current reports indicate that private owned vehicles have a utilization score of about 5%., the utilization score of CAVs would need to reach at least 20% to be profitable.
- more working flexibility if the cars will be well equipped to accommodate working while travelling to/back from office.
While the adoption of new transport modes is always exciting and attractive, various and challenging new problems will rise which need to be addressed in time before shifting towards a large-scale adoption of CAVs.
Research Science will definitely need to play a role in evaluating the impact and benefits of such new transportation modes and also find new innovative methods to improve current traffic conditions in various settings around the world.
Our Team Member's Interview on International Women's Day 2018
Our team member, Dr. Adriana Simona MIHAITA's interview on International Women's Day 2018.
UNSW-CSIRO Industry PhD Pilot Program
Australia has strong research capabilities, but relative to other OECD countries faces a number of barriers in translating research excellence into innovative products and services . Australia also has a low number of researchers in industry when compared with similar countries , declining rates of STEM education participation , and limited opportunities for industry-focused research training.
This scenario prompted CSIRO, in collaboration with UNSW, to design a new PhD program to deliver work-ready, industry-focussed researchers to bridge the research-industry divide. The first pilot program involves 9 industry partners, who in collaboration with UNSW and CSIRO, offered 11 potential research projects in the areas of data-centric society, manufacturing, materials and energy. The selection process for the pilot program ran between August to November this year, attracting over 100 high quality applicants from diverse backgrounds.
The CSIRO Industry PhD (iPhD) program aims to coordinate and bring together universities, businesses and CSIRO to jointly train new researchers to solve real-world challenges and to generate tangible benefits for industry and the national economy . We seek to help build the ‘next-gen’ workforce – to create research leaders who will understand the needs of industry and enable collaboration across the innovation ecosystem.
Industry PhD projects are closely related to Industry partners’ demands and interests. Each project involves a four-way partnership between the PhD candidate, a business, a university and CSIRO, leveraging the capabilities and experience of each partner. We seek to scale up disruption of the national research training system by involving industry throughout the development and delivery of the PhD program.
PhD candidates will collaborate with their industry partner throughout the program to understand and solve a designated industry challenge, as well as do a six-month internship. They will receive industry-relevant training, including leadership, entrepreneurship and innovation. CSIRO and the universities will provide access to world leading researchers and infrastructure. Candidates’ engagement with industry partners will significantly improve their employability and capacity to contribute to the innovation ecosystem.
The response to the pilot program call for applications exceeded our initial expectations considering it was restricted to domestic applicants only. The successful recruitment campaign demonstrates significant interest in industry-research interaction with applicants declaring that “it is different to every other PhD degree out there, working side by side with industry”, and “this scholarship offers a very unique combination of industry and research experience”.
As we further develop the CSIRO iPhD program model and identify necessary resources, we will look to scale it to include more university and industry partners. We believe that every interaction between the research-industry divide is a critical move towards generational cultural change and securing Australia’s all-important innovation pipeline.
To apply for the CSIRO iPhD program, stay tuned for next rounds to be announced through the program’s web site at http://www.csiro.au/industry-phd. For more information about the program, contact Dr. Fernando Koch (firstname.lastname@example.org), Program Manager.
Some big news from our team
- Dr Chen Cai appointed as Board Director of ITS Australia
- Data61 has a 10-year membership
- Dr Wen Tao 1 of 3 finalists for the Young Professional Award
Data61 Team at the 24th Intelligent Transportation Systems World Congress
24th Intelligent Transportation Systems World Congress
DATA61|CSIRO will be participating at the ITS World Congress in Montreal from October 29th to 2nd of November 2017. Data61 is shaping Australia’s data-driven future and aims to be one of the top leading data-focused research and innovation powerhouse. This year, the Advanced Data Analytics in Transport (ADAIT) group in Data61 will be present at Stand 204 and showcase its capability on: smart transport analytics, traffic simulation and connected vehicle investigations. The team will demonstrate its cloud hosted real-time transport data analytic platform with the following functionalities: data-fusion from multiple real-time transport data sources, monitoring of real-time traffic conditions, visualising mode choice and routing behaviour of public transport users in Sydney, Australia, while also offering the possibility to run macroscopic large scale traffic scenarios for traffic simulation purposes. Connected vehicles are also a focus point for the team which works on analysing GPS positioning accuracy from connected heavy and light vehicles using the DSRC technology. ADAIT will be also hosting 2 special interest sessions on “On-Demand Passenger Transport: Innovative Operation Models” (SIS15) and “Evaluation of Connected and Autonomous Vehicle Trials” (SIS109).
Data61 PhD scholarship round for Semester 2, 2017
The Data61 PhD scholarship round for Semester 2, 2017 is open for students enrolled or enrolling in participating universities. Applications are sought from new and existing high quality students working in Data61’s research areas of interest.
Staff interested in engaging PhD students should familiarise themselves with the Scholarship Guidelines and seek out applications from suitably qualified, high achieving students. Applications close 30 April 2017; please see the Scholarship Program website for specific details and contact email@example.com if you have any questions.
PhD Topics proposed by the ADAIT team in Data61|CSIRO:
"Impact of connected and autonomous vehicles (CAV/AV) on traffic demand and driver behaviour”, Simona Mihaita
In an era of rapid technological changes, connected and autonomous vehicles seem to be the perfect solution for dealing with challenging problems such as congestion, pollution and space optimisation in urban areas. While significant efforts are put together for dealing with regulations, standard adoption and trial cases, few studies have concentrated on studying the impact of adopting such novel technology on the driver behaviour and the actual benefit that it might provide in terms of traffic demand/travel time/congestion management. The main focus is to find innovative research ideas for assessing the benefits of CAV/AV by making use, for example, of current existing CITS platforms in Data61 .
 Mihaita Adriana Simona, Tyler Paul, Menon Aditya, Wen Tao, Ou Yuming, Cai Chen, Chen Fang, "An investigation of positioning accuracy transmitted by connected heavy vehicles using DSRC", Transportation Research Board 96th Annual Meeting, Washington D.C., January 8-12, 2017.
"Neural networks and deep learning – applications to transport modelling", Young Lee
Research on using neural networks and deep learning in modelling time series has produced mixed results. Whilst neural networks are known to possess great potential due to its ‘universal approximators’ property , their ample flexibility can lead to inference issues. Recent papers have attempted to better understand the properties of these models. A good direction in research would be to build on current results by examining some aspects of forecasting using neural networks and deep learning, including the behaviour of time series generated from a model with known parameters. Special attention should be paid to the pitfalls and potential remedies for avoiding them. The applications are enormous, an example could be in the form of time-series approaches for modelling demand for transportation.
 Hornik, K.; Stinchcombe, M.; White, H. (1989). Multilayer feedforward networks are universal approximators " Neural Networks. 2(5) pp. 359-366.
"Citywide Crowd Flow Prediction", Weihong Wang
Predicting the flow of crowds in each region of a city is of great importance to traffic management and public safety, and is very challenging due to many complex factors, such as inter-region traffic, events, and weather. Deep-learning-based approaches can be used to forecast the inflow and outflow of crowds in the regions of a city. Specifically, an end-to-end structure of ST-ResNet  can be used to work on unique properties of spatio-temporal data and the temporal closeness, period, and trend properties of crowd traffic can be modelled.
 Zhang, J.; Zheng Y.; Qi, D. Deep Spatio-Temporal Residual Networks for Citywide Crowd Flows Prediction. AAAI 2017.
"Modelling multi-modal transport systems in transition to Smart Mobility", Le Minh Kieu
New transportation solutions has led to very rapid changes in cities around the world, including the development of shared mobility, mobility-as-a-service and automated vehicles. These new technologies render significant opportunities, but also blurring the distinction between public and private transport, as well as promoting a new mode of transport-the Smart Mobility. It presents substantial challenges in modelling the evolving multimodal transport systems which has not been addressed in literature. This PhD studentship focuses on developing new mathematical models to replicate and predict the dynamics of near-future multimodal transport systems using emerging data sources and advanced machine learning techniques.
23rd World Congress on Intelligent Transport Systems
DATA61 | CSIRO will be participating at the ITS World Congress in Melbourne from October 10th - 14th, 2016. Data61 is creating our data-driven future. By combining the expertise of the CSIRO Digital Productivity and NICTA teams, we are building the world's leading data focused research and innovation powerhouse. This year, Data61 will present at Stand 3118 its capability on transport analytics, intelligent logistics and asset management. The Advanced Data Analytics in Transport (ADAIT) team will demonstrate its cloud hosted real-time transport data analytic platform. Connected vehicles are also a focus point for the team which is working on analysing GPS positioning accuracy from connected heavy and light vehicles using the DSRC technology. The intelligent logistic team will showcase the innovative solutions to logistics at Port Botany in Sydney. The asset management will demonstrate the structure health monitoring project for Sydney Harbour Bridge. The platform ingests data from multiple real-time transport data sources, monitors real-time traffic condition, and detects real-time traffic anomalies from historical traffic patterns.
Premier's Innovation Initiative - Congestion Data61 out to bust Sydney congestion