Researchers from Cornell University and Stanford trained a system to read a human's behavior and body language that precede what they would do next behind the wheel. It was trained using the machine learning algorithms and deep learning, and it was able to predict when the driver was about to change lanes with over 90% accuracy. The project, Brain4Cars, has recently used data from GPS, the attached video cameras, and information from the car's computer system.
The number of fatal accidents on the road is on the rise, and this system not only can make car systems "smarter," but it can greatly contribute to road safety for drivers. The current study for Brain4Cars is deemed "promising," but is still needed to be further tested in the real world. Product design expert Don Norman says that "many factors may change the results when applied to real people driving real cars in real traffic."