Fantine Huot

Tracking cars using telecom fiber

With the development of autonomous vehicles, there is an increasing need for sensors detecting motion in the cameras' dead angles, especially in urban areas.

At Stanford University, I demonstrated that it is possible to repurpose telecom fiber-optic cables — the same as for high-speed internet — to track vehicle traffic. This is done by sending a laser pulse through the fiber and measuring the backscattered energy. As moving objects make the fiber underground vibrate, it shows up in the recorded data.

To extract this data's full value, I developed new signal processing algorithms for vehicle tracking, combining high-performance computing and deep learning. Since the data are solely based on vibrations, this process is entirely anonymous.

Python C++ TensorFlow

Finding what you need when you do not know what you need

In Japan, I worked on the digital rebranding of Valeo, an automotive supplier company. During a market study, some customers mentioned not finding the product information they needed on the website. However, when I asked them, these same customers knew how to access and browse the product catalog.

Why did they feel like they could not find the product information? The actual problem was not that they could not find the data, but that they did not know what to do with the data.

I set out to create a tool to help customers identify the best Valeo product for their needs. Gathering input from the engineering team and the marketing team, I built a web-based performance calculator that determines the best product as a function of the customer's vehicle specifications. The customer supplies details about their desired use-case, and the web-app calculates which components in the catalog meet their requirements.

CSS/HTML PHP/SQL JavaScript