Detractors often like to bring up the potential situations where self-driving technologies could occasionally perform worse than human drivers, as a means of arguing against the deployment of these technologies. Obviously, this line of reasoning ignores the point that self-driving technologies are already less likely to cause an accident in a great number of situations/scenarios.
When taken together, it seems a reasonable estimation that future self-driving technologies will notably slash auto-accident deaths (probably mostly by taking the distracted, drug-impaired, smartphone-obsessed, very old, etc., out of the pilot seat). That’s not to say that self-driving technologies will be perfect, and they certainly aren’t so now.
Case in point, here’s a recent story shared on the GM-Volt.com forum (by “Sean-a-Tron“) involving a Chevy Volt’s collision warning system confusing the sun for an object that could be collided with.
“My Volt saved my life! While I was driving home today, going uphill; I didn’t realize how close to the sun I’ve come! Suddenly, my Volt slowed down and alerted me that I was about to hit something! If it didn’t stop me, I would’ve certainly collided with the sun. In reality. I heard a beep, felt my car slow down and thought: ‘Ah ****, my car is having issues and I’m about 2hrs away from home’. It was sort of a relief, and annoyance, that it was a false-positive on the collision warning. First time that happened to me. Almost certainly the glare of sun blinded the forward camera, causing it to malfunction.”
Sean-a-Tron shared a video of the occurrence as well. The sun did look pretty daunting in that.
Although the Volt’s autonomous collision avoidance system did seemingly “malfunction,” it did so in a way that was not dangerous, it should be noted. Better to have a false positive than a false negative.
How can you say stopping for no apparent reason is not dangerous? It is a good way to get rear ended. I think they need to be perfected before releasing auto braking features.
How many times have you had to brake for a driver, mentally (or otherwise,) cursing them for stopping with no apparent reason, only to find out soon afterwards that they *did* have a reason you just didn’t see at the time? If auto-braking saves 10 people from accidents that they could have avoided had they been more attentive, while causing 1 due to a driver making assumptions that it’ll be obvious when someone will stop, is that justified?
On the flip side, is it a good idea to add functionality that can allow drivers to be less careful, while introducing additional hazards, however unlikely, from other drivers who are less careful?
I foresee problems in the future when auto-protect features are more prevalent but not universal, and someone who’s used to those features drives a vehicle without them.