Many collision avoidance systems watch out for other cars or pedestrians to keep you safe. But this new one called Brains4Cars being developed by Cornell and Stanford University researchers adds a camera that monitors you (or the driver's, if it's someone else) body language, as well. The computer that's watching you on cam can detect your face and head movements to find cues on whether you're turning or changing lanes. With data from a radar and another camera keeping an eye on the environment, the system can warn you if it's too dangerous to turn.
For instance, if you're turning left, the left side of the steering wheel or seat can vibrate as a warning -- the researchers believe sound and visual signals could be incorporated into the system, as well. In addition, the system can also pull GPS info and issue a notification if you've taken the wrong turn, or if you're driving the wrong way down a one-way street.
To develop the system's algorithm, the researchers recorded 10 people's driving activities for two months. They then used the data to analyze a separate set of driving videos. The result? Their algorithm was able to predict the driver's actions 3.53 seconds in advance and got it right 77.4 percent of the time. Obviously, it's nowhere near perfect yet -- shadows seem to confuse the system's face detector, and it can't predict turns when the driver doesn't move his head to peek at incoming traffic.
The researchers are working on it, though, and are also looking to add more features including tactile sensors on the steering wheel and brake pedal, as well as infrared sensors. They're also considering the ability to detect whether you're looking at a phone or smartwatch to warn you against texting and driving at the same time.