Rational design of Floating Car Data collection applied to a wide road network, production of dynamic speed indicators from data sources resulting from C-ITS services and data fusion
- Level of qualifications required: PhD degree in Machine Learning / Data Science / Data Fusion / Data Assimilation
- Main field research field: data analysis, machine learning, non-linear regression, statistic, transfer learning, data fusion, Big Data, Artificial Intelligence, traffic modelling, intelligent transportation system
- Starting date: from December 2022 (subject to discussion)
- Duration of contract: 1 year and half (18 months + possible extension)
- Deadline to apply: October 2022
- Main location: Lyon, France (LICIT-ECO7, UMR UNIVERSITÉ GUSTAVE EIFFEL – ENTPE)
- Project team:
- Pierre-Antoine LAHAROTTE (research fellow), LICIT, University of Lyon, ENTPE, Université Gustave Eiffel – COSYS
- Nour-Eddin EL FAOUZI (Lab Director), LICIT, University of Lyon, ENTPE, Université Gustave Eiffel – COSYS
We are looking for an enthusiastic postdoctoral/engineer candidate to carry out research in the context of a research project funded by the French State (Department of Transport). The goal of the project is to make use of various sources of Floating Car Data to produce dynamical speed indicators on a nationwide road network, with options to take into consideration the current context (usual traffic conditions, road works, etc). The assessed speed indicators are dedicated (i) to integrate the route planning algorithm of the smartphone app Coopits  developed by the French Transport Department and (ii) to be confronted to the TIPI database (feeding Bison Futé’s prediction). The LICIT-ECO7 and the French Department of Transport have an agreement to work on the stream of data generated by the smartphone app Coopits providing C-ITS services to end-users. The collection of log data is stored on a SQL server within a relational database. In parallel, the French Department of Transport intends to invest in the purchase of Floating Car Data (FCD) to produce reliable daily profiles of traffic indicators on the main part of the nationwide road network.
The objectives of the current project are 4 fields:
- Investigation of the options available to rationally designing the purchase of FCD datasets across the nationwide network and supporting the French State to write a specifications’ note. This note must be included in the specifications document to purchase an historical dataset of Floating Car Data.
Especially, the objective is to design an algorithm as a decision-making aid and aiming to reply to the question: what duration and part (cities) of the road network should be targeted in order to collect enough data to reproduce traffic trends on the full network (nationwide scale)?
This first task should be based on a road network built as a reference for next tasks.
- Analysis of the data collected through Coopits smartphone app, production of traffic indicators such as Space Mean Speed and assessment of the statistical relevancy of the traffic indicators built from the raw data. The production of traffic indicators from Coopits data is not only targeting an off-line purpose to achieve typical profiles for traffic, but aims to perform on-line as well to display up-to-date travel time on the smartphone app. The aggregation period for the on-line assessment should not exceed a few minutes (max. 15 min).
- Analysis of the Floating Car Data collected according to the pre-defined specifications (rational design) and production of daily typical profiles for traffic indicators (speed, partial flow volume etc) on the full road network. Based on the purchased dataset, a process in 2 steps is required:
- The map-matching on the reference road network and the analysis of the purchased dataset to build daily typical speed profiles for one part of the targeted network.
- The development and implementation of a methodology to transfer the knowledge acquired by purchasing data on some spatial location to unknown locations, i.e. spatial zones outside the range of the purchased dataset. This step requires:
- to finely feature the daily speed profiles associated to road sections by identifying relevant explaining factors ;
- to develop a methodology based on transfer learning  and/or machine learning regressorsto compute typical daily speed profiles for unexplored road sections with resilience to sparse input data. The main idea is to spatially infer the speed profiles for any new road section featured by its nearest POI and sections in its neighbourhood.
To deal with the large amount of data, it might be expected to apply Big Data approaches.
- Study the compliance between the developed traffic indicators and the ones resulting from usual tools of the French State (among others TIPI database). The objective lies in designing a database to support the developed indicators and designing the protocols to project the data on other segmentations of the road network (e.g. TIPI’s network). The application of fusion methods is an option to explore while comparing to the TIPI’s database.
The postdoc will have the opportunity to extend his machine learning, Big Data analysis and Data Fusion skills by applying it to concrete transportation problems.
The main tasks and expected contributions from the postdoc are the following:
- produce codes, analysis, contributions and deliverables to the full project, as a whole. The main contributions of the postdoc should especially focus on the following dimensions:
- the investigation of available options to rationally design the nationwide FCD data collection ;
- the development of a methodology to transfer the knowledge gathered by purchasing a Floating Car dataset on a specific location to unexplored locations (co-development of this component with a postdoc researcher) ;
- the application and deployment of the previously developed methodology to large scale network (nationwide scale)
- the analysis of historical speed indicators resulting from TIPI dataset (data collected by the French Transport Department thanks to loop sensors located along some road sections)
- the development of a statistical interpolation process (kriging, for instance) and/or data fusion process to produce a relevant and reinforced historical daily speed profile from the historical TIPI’s dataset and the historical FCD’s dataset.
- make a monthly progress report (few pages maximum) to be shared with the French Department of Transport;
- prepare materials and attend meetings with the French Department of Transport;
- contribute to the final deliverables and write research papers in journal and/or conferences.
It is expected that the successful candidate will contribute to top-tier transportation conferences and journals (IEEE Intelligent Transportation Systems, IEEE ITSC, Transportmetrica B, Transportation Research Board, Transportation Research Records, Transportation Research Part …)
At the end of the current engagement, an extension could occur to deal with the opportunity to feature the speeds collected through the Coopits log data with contextual information (usual traffic conditions, road works, etc) and to deal with the issue of assessing the compliance with the other tools used by the French Department of Transport.
The postdoc will closely interact with the LICIT’s members, members of the French Department of Transport and some of his collaborators within the C-ROADS/INDID project since the smartphone app Coopits is developed in the context of these European projects.
The postdoc will join a team composed of an other research postdoc working on similar topic with some specific skills in map-matching and hidden markov models.
We look for strongly motivated candidates with a solid background in computer science, mathematics and probability / statistics. The candidate should have good skills in programming, computer science and/or machine learning, statistics, data analysis, management of Big Data. Some work experience with Big Data libraries would be an asset.
Programming skills with Python are required.
Proven written and verbal communication skills with fluency in written and spoken English and/or French.
- subsidized catering services
- partially reimbursed public transport
- social security
- flexible working hours
- access to sport facilities
The wage is set according to the usual civil servant grid. It takes into consideration the work experience.
- 2,600-2,900€ (gross monthly salary, depending on years of work experience)
- 2,000-2,300€ (net salary, depending on years of work experience)
The application must be sent to the following contacts:
- Pierre-Antoine LAHAROTTE – firstname.lastname@example.org
- Nour-Eddin EL FAOUZI – email@example.com
- A curriculum vitae;
- The complete record of master grades (academic scores for French applicants);
- The PhD thesis manuscript;
- A short (2 pages max) motivation letter discussing how the candidate’s background and research interest relate to proposed subject and bibliographic references.
About us, the LICIT-ECO7 (shortname for Transport and Traffic Engineering Lab) – Université Gustave Eiffel & ENTPE
The Transport and Traffic Engineering Laboratory (LICIT) is a Joint Research Unit under the dual administrative supervision of the UNIVERSITÉ GUSTAVE EIFFEL and the National Post-Graduate School of Public Civil Engineering (ENTPE). It is recognized for its work in traffic modelling and engineering. The laboratory has already developed many successful applications for both traffic information and simulation tools (e.g. SymuVia).
The UNIVERSITÉ GUSTAVE EIFFEL is a state-financed scientific and technological institute under the supervision of the Ministry of Research and the Ministry of Transport. The Institute’s activities include various fields as acoustics, mechanics, mathematics, computer science, electronics and electro-technical sciences. The diversity of the approach used to carry out the different research programs gives a multidisciplinary characteristic to the UNIVERSITÉ GUSTAVE EIFFEL research teams.
The UNIVERSITÉ GUSTAVE EIFFEL research program covers many aspects of work involved within this project including driving aids, information, assistance and automation; transport networks and services; sustainability, environment and road safety. UNIVERSITÉ GUSTAVE EIFFEL has been and still is heavily involved in various projects, especially European Commission-supported research like the INTRO project. UNIVERSITÉ GUSTAVE EIFFEL’s experience will be used to assist in achieving the aims of this project.
 Lin, B. Y., Xu, F. F., Liao, E. Q., & Zhu, K. Q. (2018, June). Transfer learning for traffic speed prediction: A preliminary study. In Workshops at the Thirty-Second AAAI Conference on Artificial Intelligence.