The car will learn to understand the emotions of drivers

Cause irritability, which is expressed in aggression can be both psychological and physical fatigue. It is a dangerous source of distraction of a driver on the road. Federal Polytechnic school of Lausanne has developed the onboard system for emotion recognition by facial expression. Tests on the prototype showed that this idea could have a very interesting perspective.

Modern technologies allow to read face, identifying seven universal for all people of emotions, such as fear, anger, joy, sadness, disgust, surprise and contempt. This knowledge can find application in video games, medicine, advertising and, possibly, cars. As you know, in addition to fatigue other risk factor is the emotional state of the driver.

How to measure the emotions of a driver

I must admit that measuring emotions is not easy, especially in a contactless manner. Specialists of the research laboratory of signal processing adapted a facial recognition system for use in cars. The device relies on an infrared camera placed behind the steering wheel.

The main problem faced by developers is to force the device to recognize irritation on the facial changes of the skin. Each driver is expressed in different ways- one twitch of the facial muscles, the second one swears, the third is trying to hide emotions behind a straight face. To simplify the task at this stage of the project, the researchers decided to track only two feelings: anger and disgust.

The study had two phases. First, the system “learned” to identify emotions in a series of photos of people with different facial expressions, then trained on the same emotional videos. Training materials reflect real-life situations, including in the car. The execution speed of map operations read the emotions of samples was dependent on the used analysis methods. But overall the system worked well and in most cases with an accuracy determined by the irritability.

What went wrong?

Not without failures tests. But most often it is happened due to highly variable status of the individual. You have to understand that the complexity of the problem lies in the emotional individuality.

See also — Tesla Cars will be less likely to cause accidents in the oncoming lane

The researchers plan to conduct additional research that will improve the system, adding more advanced algorithms and human-machine interface. Emotion recognition is only a small part of efforts aimed at improving safety and driving comfort. Now in modern vehicles there is a system monitoring driver fatigue: it monitors the physical condition of the driver and if detects certain deviations, warns about the need to stop and rest. Such systems are built on the control of the driver, control the vehicle and controlling the driver’s view. For example, if the car starts to wag from side to side, the system signals the need to stop.

Advertisements