WavBrain
WavBrain has developed a device measuring just a few square centimetres that can detect drowsiness in drivers using an artificial intelligence (AI) algorithm that analyses brain activity.
Transfer project
The AI technology developed by the researchers has made it possible, for example, to use a simple single-electrode EEG headset to detect drowsiness (hypovigilance) at an early stage, all without access to a cloud server.
The maturation project consisted of conducting a wide-ranging test campaign, consolidating the technical framework, and supporting the creation of the start-up, in particular by finding an experienced CEO.
Development
The technology transferred was developed by the Institut de Recherche en Informatique de Toulouse. (IRIT)
The company
Created in 2023, the company offers technological building blocks for the early detection of drowsiness (hypovigilance) via a lightweight, low-cost EEG headset.
It is also developing a software platform to facilitate the development of embedded AI applications.
Innovation
This innovative tool uses AI code developed by Wavbrain to interpret electroencephalographic (EEG) signals in an energy and data-efficient manner. For example, it can be integrated into small microcontrollers and fitted to a headband or headrest.
In initial tests, the device detected stages of drowsiness in 99% of cases, a much better performance than current systems based on head movements or brisk steering wheel movements.
"Drowsiness is the main source of car accidents. Legislation requires new vehicles to be fitted with drowsiness detection systems. Current tools are based on head movements, eye blinking and steering wheel movements. These detections take place at a late stage of drowsiness and prevent few accidents."
Promising applications in other sectors
In addition to the automotive sector, this embedded AI code is attracting interest from the construction industry to detect fatigue in machine operators, and from the aeronautical industry to equip pilots' helmets. Jean-Pascal Aubert, Co-founder and CEO of Wavbrain, also mentions potential applications in secure connection to IT systems and the diagnosis of sleep apnoea.
Wavbrain aims to industrialise embedded AI.
Following this initial success, Wavbrain aims to create a generative AI tool capable of producing new codes adapted to embedded objects from previous codes. To achieve this, the company has formed a consortium with several engineering schools to ensure that generative AI has sufficient feedback data from as many users as possible.