Japan's Mitsubishi - in vehicle camera5 Apr 2026 08:02
OSAKA -- Major Japanese electronics maker Mitsubishi Electric Corp. has developed artificial intelligence (AI) that determines whether a driver is intoxicated by facial images and other signs of alcohol intoxication captured by an in-vehicle camera.
The company aims to work with automakers to develop the technology, so that vehicle control functions such as displaying warnings or stopping the vehicle can be carried out when the AI detects that a driver is drunk based on footage, with the goal of early commercialization.
According to Japan's National Police Agency, there were 125 fatal accidents caused by drunken driving in 2025, a decrease of 15 from the previous year. However, the fatal accident rate for drunken driving was about 7.1 times higher than for sober driving.
In the United States, the National Highway Traffic Safety Administration reported that more than 12,000 people died in drunken driving accidents in 2023. Getting behind the wheel after drinking has become a social issue both domestically and internationally, and discussions in Europe and the U.S. are progressing toward mandating the implementation of drunken driving prevention technology in new vehicles.
Mitsubishi Electric's Advanced Applied Development Center in Hyogo Prefecture commercialized a "Driver Monitoring System (DMS)" that uses in-car cameras to recognize a driver's face and detect inattentive or drowsy driving, triggering a warning sound. This DMS has been adapted to detect intoxication in 2018,
The system employs infrared cameras capable of nighttime use to capture changes in skin complexion due to blood vessel dilation, while the AI accurately measures increase in pulse rates. It estimates drowsiness from eye movements and other indicators. Additionally, the AI determines intoxication by combining vehicle control information, such as abrupt steering or braking, with other data.
To move toward practical application, Mitsubishi Electric conducted joint research with Oakland University in the U.S. in 2024 and 2025, collecting driving data from about 100 individuals of different races, skin colors, ages and sex while in a state of intoxication.
A representative of the Advanced Applied Development Center emphasized, "By detecting intoxication with high accuracy, continuous monitoring becomes possible, which can contribute to reducing accidents caused by drunken driving."
(Japanese original by Yin Tanaka, Osaka Lifestyle, Science & Environment News Department)