Interior camera uses AI to detect driver distraction and fatigue

December 18, 2019 //By Christoph Hammerschmidt
Interior camera uses AI to detect driver distraction and fatigue
Microsleep, distraction, an unfastened seat belt - many things that happen in a vehicle can have far-reaching consequences. In order to avoid critical driving situations and possibly accidents, cars with their sensors will in future not only have to watch the road, but also the driver and passengers. For this purpose, Bosch has developed an camera-based interior monitoring system that uses artificial intelligence (AI).

The system is scheduled to go into series production in 2022. Then the safety technology, which warns the driver in the event of fatigue or distraction, for example, is to become the standard in new vehicles in the European Union. The EU Commission expects that its new vehicle safety requirements will save more than 25,000 lives by 2038 and prevent at least 140,000 serious injuries.

Looking inside a vehicle should also solve a fundamental problem of self-driving cars in the future. For example, if the system is to return the responsibility for driving to the driver after a long journey in automatic mode, it must be ensured that the driver neither sleeps nor reads the newspaper or e-mails on the smartphone.

That's why Bosch is developing an interior monitoring system that recognizes the hazard of distracted or tired drivers and can both warn and support them. A camera built into the steering wheel detects when the driver shows signs of fatigue, when he is distracted, or when he turns his gaze toward the side or toward the rear seats. The AI algorithm evaluates the driver's head posture, viewing direction and movements and draws the right conclusions. It warns the driver in case of carelessness, recommends breaks when he gets tired, or even reduces the speed of the vehicle - depending on the wishes of the vehicle manufacturer or legal requirements.

The developers used sophisticated image processing algorithms and machine learning to teach the system to understand what people are doing in the driver's seat. Take fatigue, for example: the system is trained by taking pictures of real driving situations and learns how tired the driver really is by means of camera shots with eye opening and eyelash frequency. It can therefore provide customized warnings in any situation or support the driver with the help of driver assistance systems.


Vous êtes certain ?

Si vous désactivez les cookies, vous ne pouvez plus naviguer sur le site.

Vous allez être rediriger vers Google.