Panasonic to release artificial intelligence drowsiness driving system

1. Measure blinking, expression, etc. in a non-contact manner, detect unaware and shallow drowsiness

When people become drowsy, various signs such as "sleepy expression" and "how to blink" appear. By measuring these, you can detect drowsiness.

We have newly developed a technology to detect blinks, facial expressions etc from the camera image with high precision in a noncontact manner by utilizing the image recognition technology cultivated in the development of surveillance cameras and the like. Furthermore, various measurement results related to drowsiness, blinking, expression and so on gathered so far are made into a database and analyzed from the physiological point of view about the relationship between about 1,800 parameters concerning blinking, expression, etc. and drowsiness. Also, based on the analysis result of drowsiness expression collaborated with Ohara Memorial Institute of Labor Sciences, we developed a unique AI to estimate drowsiness level.

This enables you to detect unconscious, shallow drowsiness and classify the level of drowsiness you have detected.

Extract (contour) the outline of the eyes and monitor the time change of blinking from the degree of opening of the eyelids (graph).

2. Measure the passenger compartment environment and predict the subsequent trend of drowsiness

Generally, it is said that sleepiness depends on the in-vehicle environment such as temperature and brightness, because people are less likely to become sleepy in cold and bright environments, and they become sleepy in warm and dimly lit environments. However, it was difficult to estimate the susceptibility of people to sleepiness only from the ambient temperature, such as people with thick clothes and thin clothes under the same temperature environment.

We have revealed in collaboration with Chiba University that, regardless of clothing condition, heat release from the human body is related to drowsiness after a certain period of time. In addition, we developed a technology to contactlessly measure the amount of heat dissipated from the human body using our own infrared array sensor "Grid-EYE (Grid eye)" [2] . In addition, we clarified the ambient brightness measured by environmental sensors and the influence of time lapse on drowsiness.

With these, it became possible to predict how current drowsiness will change from the amount of heat radiation from the body of a person measured non-contact and the surrounding brightness.

3. Measure human's thermal sensation and thermal comfort, maintain comfortable arousal state comfortably

Controlling the temperature and air volume according to the predicted drowsiness level makes it easier to maintain the arousal state, but if the temperature is too low, people feel cold and thermal comfort will be compromised. We applied the knowledge of the thermal environment and physiology cultivated so far in the development of room air conditioners, etc. In collaborative research with Nara Women's University, we can apply thermal insulation in the environment such as the interior of the cabin, We developed a sense estimation technology * 2 . By constantly monitoring the thermal comfort of people with the infrared array sensor "Grid-EYE", by selecting the most suitable means such as air conditioning, it is possible to maintain the awake state comfortably * 3 .

2 Technique to contact human's thermal sensation (hot and cold feeling) noncontactly by using infrared array sensor
3 If it is detected that the current drowsiness level of the driver is high, you will need alerts and break instructions with voice etc.
<Estimation of human thermal sensation using infrared array sensor image (under same environment)>

【Conceptual diagram of drowsiness control technology】

【Cockpit mounted image】

【Terminology】

[1] 5 levels of drowsiness Equally spaced scale

The value of drowsiness level evaluated from facial expression. 1. Slightly not sleepy, 2. Somewhat sleepy, 3. sleepy, 4. quite sleepy, 5. very sleepy, classified in 5 stages.

Source: Kitajima Yoki, Numata Nakaho, Yamamoto Keiichi, Goto Mihiro: "A Study on Prediction Method of Sleepiness During Driving (1st Report, Effective Methods for Estimating Sleepiness Expression and Predicting Sleepiness Changes)"

Proceedings of the Japan Society of Mechanical Engineers (Part C) 63, 613, PP 93-100, 1997

[2] Infrared Array Sensor "Grid-EYE (Grid Eye)"

The infrared array sensor is a sensor in which pixels for detecting infrared rays of 10 um band radiated by people and objects are two-dimensionally arranged, and a two-dimensional temperature distribution can be measured. Our infrared array sensor "Grid-EYE" is a 64 pixel sensor, but by mounting a high resolution algorithm and swinging at an optimum angle and speed, a clear thermal image equivalent to about 7800 pixels can be obtained .

Your voting gives me strength. Thank you.