Cognitive sciences and AI

Jean-Baptiste Guignard - an audio-digital engineer with a PhD in cognitive sciences, who did research at the UTC Costech Lab for 10 years – created his start-up, HINS, in 2015. The company implements and commercializes a solution to recognition of hand movements as a concrete application of artificial intelligence (AI) based on biomimetics.

Cognitive sciences and AI

During his time as Professor at UTC, Jean-Baptiste Guignard came up with an original idea: inventing a system to enable a composer to make real-time changes to a piece of music by moving his/her hands in front of a smartphone’s camera. Making use of open source solutions and a signal processing software package, J-B. Guignard was able to assemble and test a prototype which he presented at a general public TV show.

This in turn allowed him, albeit fortuitously, to get to know the Renault-Nissan. The automobile manufacturer has shown interest in this gesture recognition technology that does not need any specific hardware to operate and to control certain functions installed on board its cars. Consequently, our “lab. and lecture-hall” specialist decided to try his hand at entrepreneurship. “I have to do loads of work and the risks are quite high, but in France fortunately we can access various portals such as “Jeune Entreprise Innovante” and the Government incentive R&D tax rebate programme, which are unique to the country.” After three years, HINS with 31 staff and its home office in Los Angeles now has taken on an international dimension. HINS’s customers are French, Americans and Chinese.

Possible applications are diverse. The start-up continues to collaborate with the automobile construction sector in France but is also present in California under another corporate name, Clay Inc., who develops technologies specific to virtual and augmented reality devices. Over the past few months, China has become a market target. “The manufacturers and operators of Chinese mobile phones are also interested, inasmuch as their clients will be able to lock their phones with a move of the hand, create and add visual effects round an image of their hand (flames, avatars, etc.). Other “apps” are in the pipeline, such as a system aimed at public works that would allow a site operative to consult his i-pad without having to take off his heavy-duty gloves (to read a map, or a lay-out, for example).

Innovation beyond the specialties

The success story of HINS, akin to the track-record of its founder, is a direct result of exchanges among biology, psychology, philosophy and engineering sciences. The clinical distance with respect to technology and its forms as introduced by use of social sciences and humanities has proven to be a driving force when it came to proposing innovative products on the market-place. With his dual background as a research scientists and his acquired skills in cognition sciences, Jean-Baptiste Guignard was able to develop an original combination of computer vision and artificial intelligence. The main challenge for Clay Inc. when it was launched was to be efficient despite the limited computational power of a classic smartphone: “The demands of our earlier versions of the code exceeded the computing levels available on smartphones at that time; now we only use 6% of their CPU processor power”. To go beyond this computational power constraint, our engineering entrepreneur explored some unexplored territory. “Top-heavy” methods that previously relied on so-called ‘deep learning’ or on self-reinforced learning patterns have simply been abandoned. “Hardliner developers tended to focus only on computer science models and statistics; at Clay Inc., we also seek inspiration in biomimetics”, explains our CEO. Among our sources of inspiration, we can cite the theory of auto-poiesis developed by two Chilean biologists, Humberto Maturana Romecín & Francisco Varela García to describe the self-preservation mechanisms that living organisms have invented and established.

For its general structure, Clay Inc. mimes the metabolism of a bacteria. The interpretation of a video stream schematically copies the way a human retina functions. In AI, recognizing shapes, a signature, an ID, a sentence all represent major challenges. The algorithms Clay Inc. implements make good use of discoveries made in cognition psychology, notably in the field of categorizations. In order to satisfactorily solve certain concrete problems such as identifying a hand where a finger or two are missing, for example. Research engineers are currently studying the strategies adopted by humans and their capacity to carry out fuzzy categorization. This approach combining as it does experimental work and industrial applications is, in essence, the DNA of the start-up HINS from its Bordeaux origins to its ambitions to conquer the world.