Self-driving cars have the potential to have a significant impact on our health and wellbeing, but not just in the obvious way you might think. One of their biggest potential selling points has always been that they could be safer than human drivers, and I’m certainly inclined to go along with that. Robots don’t get tired, they don’t get drunk, they don’t get distracted by the kids in the back; they just keep scanning and keep driving. The other way they might help is simply by freeing us up to do other things. I can’t begin to estimate the amount of time that people waste in cars every week but it’s too much. Imagine if you just programmed the sat nav and were then free to fritter away your time checking your work email or even to do something useful like read a good book.
Being good little scientists, though, what we mustn’t do is take any of this for granted, especially the safety part. We mustn’t assume that they’re safer, we have to check. Luckily for us a team from the Transportation Research Institute at the University of Michigan has done just that, unfortunately the answers aren’t what we necessarily expected.
In their report they come up with four major findings, the first of which is that robotic cars are involved in more accidents per million kilometres travelled than ones driven by people. A lot of media outlets have run with this initial outlook and used it as a stick with which to beat our automated friends but I don’t think this paints a fair picture, it’s almost like they chose to ignore the next three major findings.
The next finding is that the 95% confidence intervals on each data set overlap. This means that it could well be the case that self-driving cars are more safe than human driven ones, we simply don’t have enough data to know yet. The next finding, and this one is particularly important, I feel, is that in all the 11 crashes that robot driven cars have ever been involved in, not one of them was found to be the fault of the robot. The final finding was that the injuries and damage caused in robot accidents were less severe than in human ones. These last three points, for me, could easily outweigh the first; but the bottom line is: we need more data.
Interesting points that need to be followed up is why the robot cars have more accidents, especially as it isn’t them that’s causing them. Are nearby motorists distracted by the sensor arrays? Do the robot cars have a safe but different style of driving that humans aren’t used to? These are questions we’ll need to answer before it’s likely that robotic drivers begin lightening our load on the highways and byways.