Human error is a part of life and, unfortunately, it is a part of driving as well. People make mistakes, lose focus or react poorly in circumstances that could lead to unsafe driving and ultimately cause injuries or even fatalities in a car accident. The best way to prevent accidents so far has been to teach drivers the correct way to drive and pass laws to regulate unsafe behavior and promote careful driving.
Now, though, an invention from Cornell University’s computer science department could offer a new way to cut down on accidents and keep drivers from getting themselves into unsafe situations. An assistant professor and his team have developed a car safety system that will warn drivers if they are about to make mistakes behind the wheel. The team that thee warnings may be vibrations, lights or sounds that will alert the driver to the potential danger and urge him or her to react.
Professor Ashutosh Saxena, who led the invention team, observed that modern cars have systems in place that monitor outside factors like weather, unsafe roads and hidden obstacles. “The next leap forward” will be systems like Saxena’s Brain4Cars, which monitors internally, observing the driver and detecting whether he or she is sleepy, distracted or acting erratically.
Saxena’s team studied videos from 10 drivers and the roads on which they traveled for two months. They tracked the drivers’ movements, reactions and facial expressions in correlation to their actions while they drove. The test videos were taken over a span of 1,180 miles of both highway and urban driving.
This video was then used in a computer that monitored face detection and tracking software to identify the driver’s head movements and connect those movements to driving-specific behaviors, such as lane changes, left and right turns, stops and starts.
The tracking system that is used in Brain4Cars uses this data to anticipate possible actions and reactions that a driver could make. It constantly monitors the driver’s behavior and reports its predictions to the central safety system in the vehicle. In tests against other videos and driver data, the system used for Brain4Cars made more than 75 percent of predictions correctly and anticipated drivers’ behavior almost 4 seconds before they reacted. These extra seconds could be life-saving, Saxena noted.
Although the system is not 100 percent error-free yet, the basic model is promising. However, some of the errors the team found include face recognition software issues, specifically noting that it sometimes gets confused by shadows and plants outside the car. Also, the tracking component may get thrown off by interactions between the driver and passengers.
Still, in the future, Saxena and his team hope to incorporate more technology for improvements, including the use of 3-D cameras, infrared cameras to help nighttime driving and pressure sensors to monitor the pedals.
At Levine Law, our Denver auto accident attorneys are always looking for innovative ways to promote safe driving and keep our roads accident-free. If you have been injured in an accident as a result of someone else’s negligence or because of an automobile malfunction, you may be entitled to benefits in a personal injury claim. To discuss your case, contact Levine Law today for a free consultation.