From December 2013 to March 2017, I was a PhD student in Machine Learning. My goal was to build a complete solution for gesture recognition to control mobile robots in outdoors areas. Furthermore, I developed specific machine learning technologies suited for accurate, real-time robotic control through gestures.

I have been mainly tackling challenges related to detection and classification of multidimensional time series in streaming. This led to the development of an accelerated technique to compute distances between time series (AATLD 2015), along with ideas to analyze historical streams to compute hyperparameters and eliminate low-quality gestures in the training set (ICPR 2016).

The main tangible result of this thesis was a custom wired glove and a software pipeline for gesture training and control. After training, the user could use one hand to drive a system such as a pan-tilt camera or a mobile wheeled robot. It was designed to be integrated easily with other systems.


Conference papers


Posters (excl. papers)

Presentations (excl. papers)

  • 20-min talk in seminar "Journées de Palaiseau on Human-Computer Interfaces" [Thales event]. May 24, 2016, Palaiseau, France.


  • Gesture control with wired glove. Techno Day VIVEA [Thales Event], Jun 4, 2016, Gennevilliers, France.
  • Pan-tilt military camera gesture control. Sophie and Portable Optronics User's Club [Thales Event], Oct 27, 2016, Bordeaux, France.