The Deal With Self-Driving Cars

The+Deal+With+Self-Driving+Cars

Self-driving car technology has begun hitting the road at a pace much faster than ever expected — beating many analysts’ expectations by more than a decade. Regarding production-grade, available self-driving systems, Tesla’s (TSLA) Autopilot is by far the most readily available and mature.

A recent incident, however, has had many wondering whether Tesla’s strategy of rapid deployment has been one in the best interest of public safety. In May of 2016, a driver’s Tesla engaged in Autopilot crashed into a white trailer, immediately killing the driver of the car.

Such an incident — while incredibly tragic — should be examined in the context of a larger ecosystem, rather than be cherry-picked as an argument against self-driving car technology deployment itself. For one, the current usage agreement of Autopilot necessitates that the driver remains engaged and observant. In contrast, this driver was completely disengaged, with some reports suggesting a movie was playing in the car.

Most compellingly, a January 2017 inquiry by the National Highway Traffic Safety Administration (NHTSA) found that Tesla’s deployment of Autopilot, in fact, reduced crashes by nearly 40%. This globally contexted, data-oriented approach to examining the problem is the one that should be taken, as opposed to sensationalizing single incidents.

One could even argue that had the trailer been equipped with self-driving technology as well, it could have moved out of the way, or somehow signaled to the approaching cloud-connected Tesla that a stationary object was present.

As more automakers introduce self-driving functionality to their cars, one could extrapolate that the accident incident rate would only continue to drop. The effect would even be compounded by the fact that as the number of self-driving cars increases, the number of car-to-car interactions (edges) in the pool of cars programmed into the system that are autonomous-to-autonomous would increase at a rate faster than the number of self-driving cars (nodes) themselves.

Thus, increasing arguments for waiting to deploy self-driving technology are severely short-sighted. When examining the larger ecosystem and the more significant potential of self-driving to very quickly eliminate highway-based accidents, it becomes increasingly clear that there can only be one path going forward: one that accounts for errors in self-driving technology but still continues to recognize its utility in our modern world.