The next wave of traffic safety appears to be driverless vehicles. Already, driverless cars are being heavily tested, and driverless semi-trucks are just beginning to be tested. The initial reports of driverless vehicles appear to be good; with apparently fewer accidents than with human drivers.
Currently, driverless vehicles are equipped with onboard computers and sensors that have the capability of seeing and detecting objects at 360° at all times. This is clearly one benefit that driverless cars have over humans, who are only capable of seeing in one direction at a time.
Driverless vehicles can also take in much more information simultaneously than can humans, in large part because of the increased degree of constant monitoring. For instance, a driverless vehicle could simultaneously take into account that a vehicle is approaching rapidly from behind, that a person on the side of the street has just entered a crosswalk, and that a van in front of the vehicle has suddenly slammed on the brakes. In contrast, a human would likely miss at least one of these activities because of their inability to see in all these directions at once.
A key to the success or failure of driverless vehicles will in large part be their ability to make split-second decisions based upon all of this incoming information (assuming that such events have been properly detected). In the scenario described above, how should a driverless vehicle react? Should the brakes be applied to avoid hitting the vehicle in front, or should the vehicle swerve to avoid an accident, while at the same time also avoid the pedestrian?
In some cases, driverless vehicles will need to make decisions that will necessarily involve choosing one crash over another. As an example, a situation may arise where the vehicle will need to take evasive action and determine whether to hit a car in front of the vehicle, or swerving on a sidewalk to avoid a crash (but potentially injuring pedestrians).
The specific action that may be taken will likely be programmed into the vehicle’s computers. Thus in a very real sense it may be the programming engineers who decide which people may be the victims. Additionally, such programming may, in fact, be designed to favor the passengers in the driverless vehicle over those in other vehicles, or even pedestrians.
In such crashes, who should be liable? As driverless vehicle crashes occur, these issues will need to be addressed. Hopefully, however, if driverless vehicles do become commonplace, crashes will become a rarity.