Yesterday I went to the 4th Innovation Forum organised by the DGA French procurement agency on the campus of a renowned engineering school just outside Paris.
The Forum is the occasion for small and medium enterprises (SMEs), doctoral students and laboratories to show over 100 research projects they are working on that have all been supported technically and financially by the DGA and that will be at the heart of future defence systems. Amongst these available financial aids, RAPID, which supports SMEs and ETI (economic and technological intelligence) projects, had a pot of €50m in 2015, a 25% rise in funding over the past three years. In addition, the DGA has spent €11m to fund some 450 doctoral theses, post-doctoral internships and research abroad.
Arranged around 10 principal themes, including communications, new materials, robotics, and medicine, the doctoral students themselves, the CEOs of the SMEs, laboratory directors were all on hand to explain the “explanatory” panels that were incomprehensible to the non-scientist and non-engineer!
But I did understand most of this: the title translated into English becomes: a Bayesian statistical method to analyse hyperspectral images; I prefer to call it “Lynx Eye”. Just as a reminder: a Bayesian probability is a quantity that we assign to represent a state of knowledge or a state of belief, in other words a probability assigned to a hypothesis.
Yoann Altmann, a post-doctoral researcher at Edinburgh’s Heriot-Watt University, was kind enough to spend time explaining the “Lynx Eye” to me: a normal colour camera has three colour channels: red, green and blue. The Lynx Eye, on the other hand, can capture hundreds of colour nuances and detect an anomaly. For example, in the third row, figure 3, the blue picture clearly shows up an underground stream that is invisible in the green picture. In the third row the system has been able to discriminate between materials that are spectrally similar: the different shades of blue in figure 5 indicate a path that is invisible in the green photo (but that Google Maps had already seen!!) but also indicate different varieties of trees (the light blue patch at the top right).
The obvious interest for the defence sector is to be able to detect not only vehicles, people and objects that have been camouflaged, but also chemical agents and small differences on the ground and in vegetation that could indicate a recently trodden path or replanted section of forest.
This robot, which can quickly find the best way to continue its job even when injured, was fascinating. It is best explained by the video:
During the Forum, three researchers who defended their thesis in 2013 were rewarded by Laurent Collet-Billon, director of the DGA. Only their first names were revealed: David, for fundamental medical research which led to the understanding of how paramyxoviruses (influenza, mumps etc.) replicate thereby clearing the path to new, targeted antiviral treatments; Xavier, for work on laser beams that opens the path to the development of powerful but compact ultraviolet lasers that could detect at a distance biological or chemical substances and explosives; Lucie for her patented work on vibro-acoustics that will be used to improve the acoustic stealth of submarines.