A lecturer-researcher at the Centre of Microelectronics in Provence, and more specifically in the SFL department (Manufacturing Sciences and Logistics), Dominique Feillet presents his collaboration with the company Cityway.
Could you explain your research?
My work focuses on route-planning algorithms applied to the transport of people or goods.
What types of companies do you work with?
Over the years, I have worked with major companies such as SNCF and CMA CGM, but I am going to present our collaboration with Cityway, a company specializing in information technologies applied to mobility.
We have been collaborating with Cityway for more than 10 years. This collaboration has led to two PhD theses, and we have also just submitted a European project in which Cityway is a partner. The aim of our joint work is to improve their route-planning tools. A route planner is a reference tool for simplifying users’ journeys. It makes it possible to find the best route to get from point A to point B using one or more modes of transport (public transport, walking, bike sharing, etc.). This is a very complex problem to solve due to the difficulty of accessing all the data (from multiple stakeholders) and because each client will have specific requirements.
Do you have a concrete example of a successful collaboration with this company?
The work carried out as part of the collaboration with Cityway was the subject of a PhD thesis defended in 2017. Alexandre Iglesias, then a PhD student at CMP, worked on improving this route planner.
His thesis, “Multi-criteria route planning in multimodal transport”, aimed to develop a planner capable not of finding the single best path, but a set of good paths, to give users a choice. An optimized choice of routes that took into account the itinerary, duration, price, number of transfers, etc. This multi-criteria aspect of the problem was complex because, to optimize the algorithm, we had to model user preferences and then identify a set of sufficiently diversified routes that matched these preferences well.
Another important criterion was the planner’s response time. Users expect a very fast response when searching for a route—on the order of a second.
This work is highly representative of the expertise of the SFL department, which connects three disciplines:
- industrial engineering, for integrating decision-support tools into industrial processes
- applied mathematics, for modelling and solving an optimization problem—here, computing the best route
- and computer science, for developing an algorithm capable of solving this problem as effectively as possible, i.e., by proposing solutions of the best possible quality and as quickly as possible
