At the closing ceremony of the 2017 DataCity open innovation programme, organised by NUMA and the City of Paris, the participants presented their new urban services, which have already been tested under real-life conditions. In partnership with SUEZ and the City of Paris, the craft ai start-up has developed a service to inform citizens, companies and janitors in real time of the time when the household waste collection trucks will arrive. A close-up on this predictive technology with Dr. Mathieu Boussard and Sylvain Marchienne, project managers at craft.ai.
Could you introduce yourself and tell us about your company’s activity?
Matthieu Boussard: I work at craft ai on the development of an Artificial Intelligence engine which lies at the heart of craft ai. craft ai provides a machine learning service that detects the habits of users – in particular of mobile applications – in order to improve the performance of smart devices and to automate certain tasks.
Sylvain Marchienne: I am a data scientist at craft ai, working on varied projects. I analyse data and create models to make predictions.
Could you tell us more about the challenge you took up during the DataCity programme?
M.B.: The goal of this challenge was to reduce the time that the household waste bins spend in the street by using the data provided by the city authorities. We needed therefore to learn about the usual schedules of the waste collection trucks.
S.M.: In this project, we worked with both SUEZ and the City of Paris. Thanks to the data provided by the city authorities, we were able to geolocate the routes of the waste collection trucks. This allowed us to create models to precisely anticipate the times when a truck would arrive in a given neighbourhood. We can now precisely predict when the truck will arrive in a precise geographical zone. A trial is currently under way in the 14th arrondissement of Paris.
How did you collaborate with SUEZ?
S.M.: SUEZ provided logistical support in the form of the monservicedechets.com website. This digital platform contains predictions of the times when the waste collection trucks will arrive. We worked together with SUEZ on the presentation of these results.
M.B.: The trial has enabled us to create an innovative and lasting service. We achieved this by studying the conditions of viability of the service in detail. First, we had to make sure that this service meets a real need amongst the users. We also paid close attention to the way in which the information was made available to the users. Finally, we looked into the economic feasibility of the service.
What have your learned from this experience?
M.B.: Our approach based on artificial intelligence and our collaboration with SUEZ on this subject gave us the possibility to confirm the viability and the benefits of the service for local authorities. The users were also closely involved in the challenge right from the start, and tests will be under way until September, in order to double-check two important points : first the accuracy of the predictions of course, but also the fact that the way these predictions are made available to the users meets their expectations.