In a groundbreaking validation test, Blueskeye’s machine learning algorithms were put to the task of detecting driver drowsiness in a group of 20 participants. The test took place in an immersive driving simulator consisting of an Audi TT car surrounded by a curved screen.
The drivers engaged in a monotonous driving task for up to an hour, following a car on a motorway. At five-minute intervals, they assessed their level of tiredness using the Karolinska Sleepiness Scale.
Blueskeye’s algorithms utilized Near Infra-Red cameras mounted in the car’s left and right pillars to capture and analyze eye, head, and facial muscle movement multiple times per second. This cutting-edge technology was able to identify early signs of fatigue with remarkable accuracy.
Reports suggest that the machine assessments of tiredness met the EU’s threshold of 40% sensitivity, indicating that the system is capable of detecting when a driver is drowsy and meeting the standard required by the EU for new cars.
Professor Michel Valstar, Blueskeye’s Chief Scientific Officer, expressed his ultimate goal of utilizing AI face and voice analysis through existing vehicle cameras and microphones to aid automotive manufacturers in developing vehicles that can respond to occupants’ emotions. This technology validation for driver drowsiness and attention demonstrates the effectiveness of Blueskeye’s underlying technology, positioning them well to assist clients in meeting future legislative demands like the EU’s Euro NCAP Vision 2030.
Dr. David R Large, a Senior Research Fellow in the Human Factors Research Group, highlighted the benefits of collaboration with commercial companies like Blueskeye as an opportunity to apply academic expertise and facilities to solve real-world challenges. This collaboration reflects the ongoing evolution of the automotive industry, with a need for new technology to enhance safety and sustainability. Participants in the study responded positively to the idea of their cars monitoring them for signs of drowsiness and intervening if necessary, indicating potential receptiveness to such safety measures in the future.