The technology for tracking the movements of the hands of distracted drivers calculates how long it will take for a person to take control of the car in an emergency.
When automakers settle everythinglegal aspects of using autonomous control systems of the third level, vehicles will begin to independently deliver people to their destination. This technology will allow the driver to be distracted from the road to write a text message or watch a video.
In emergency situations, such cars will still rely on peopletherefore, it is important for them to know how quickly he can respond and take control. To solve this problem, researchers from the University of California developed a special surveillance system.
The team took an existing program fortracking the movements of the whole body, and adapted it to monitor the location of the wrists and elbows of the driver, as well as the passenger, if any. They then developed and used machine learning algorithms to teach the algorithm how to interact with unmanned control systems.
During the tests, the technology determined the positioneach of the eight key joints with an accuracy of 95%, but it failed when the driver covered parts of his body or he was not wearing typical clothes with many patterns that were not presented in the images analyzed during training.
According to the team, most of the problems can be solved by placing cameras in more convenient places and expanding the set of training photos.
However, the human reaction is still playingBosch is developing a new generation of interactive 3D displays so that drivers can quickly notice and interpret information on the dashboard.