0345 811 9595

Mon-Fri 9am to 5.30pm

Self-driving cars

...But Were Too Afraid To Ask

Because it is scary, isn't it? The sight of a car driving itself surely evokes similar feelings to those who first witnessed “powered carriages” after centuries of horses. We are entering the dawn of self-driving cars. Expect plenty of cries of “witchcraft”.

Self-driving cars bother people. They especially seem to bother those who make the sort of cars that you have to drive yourself. Indeed, an American advert for the Chrysler Dodge Charger openly compares the very concept of self-driving cars to The Matrix, arguing that this is but the first step to an awful process that will end “with robots harvesting our bodies for energy”.

But the truth is, there's very little reason to get frightened, and plenty of reasons to get excited by the prospect of self-driving cars.

Popular Science Magazine have this month run an extensive feature on Google's self-driving cars. It's fascinating and very much worth your while. However, it's quite a lengthy read, so for those who're pressed for time, allow us to give you something of a primer for self-driving cars.

How Do Self-Driving Cars Work?

Last week we reported that Tesla have entered the race to develop commercially viable self-driving cars. In that article, I incorrectly stated that self-driving cars rely upon cameras to assess the world around them. Rather, they're powered by a very sophisticated programme called Google Chauffeur.

Google Chauffeur works with a device called a lidar (light detection and ranging), which works on the same principles as radar and sonar. However, it's much more accurate, generating up to 1.3 million voxels per second.

A voxel is a bit like a three-dimensional pixel, representing a point in space rather than a point on a screen. A million voxels grouped together created a point-cloud, a supremely accurate 1:1 scale 3D model. Self-driving cars, then, truly can “see” the world around them, but because the world is viewed as data, it can be interpreted and understood.

How Safe Are Self-Driving Cars?

According to the existing data, self-driving cars are at least as safe as human-driven cars. They might even be safer.

As Google's tests are being conducted in the US, their data is being compared to US road safet statistics. In America, drivers are involved in an accident an average of once every 500,000 miles. Accidents that result in an injury occur even less frequently, once every 1.3 million miles or so. Fatalities occur once every 90 million miles.

Google's experiments have so far clocked up 500,000 miles, and have so far experienced but one accident. And even this can be attributed to human error, as the Chauffeur programme wasn't activated when the incident occurred.

So whilst the argument could be made that self-driving cars are as safe as your average human driver, Google stress that their system hasn't yet encountered enough challenging real-world situations for them to be absolutely sure. The signs, however, are looking very positive.

The Google Chauffeur programme can make mistakes. It performs hundreds of diagnostic checks a second, yet glitches currently occur about every 300 miles. Google insist, however, that the vast majority of these are nothing to worry about, though at the moment, a degree of driver participation is still required.

What's It Like To Ride In Self-Driving Cars?

The business lead of Google's project is called Anthony Levandowski, and he's been relying on self-driving cars for his daily commute for some time now. Thanks to this technology, his input is required for just 14 minutes of his hour long, 45 mile commute.

Because the technology still isn't perfect, self-driving cars still depend upon human input at key moments - such as when a collision seems imminent, or whilst navigating areas where pedestrians could unexpectedly run into the road (such as the suburbs).

But apart from that, all it takes is the touch of a button to switch the car from “Manual” to “Auto Driving”, at which point the driver can relax. Google Chauffeur and the lidar takes care of the rest.

According to the current data, self-driving cars can travel some 36,000 miles before making a mistake so severe that human intervention is required, and even in such instances it's more likely to be a case of the technology misinterpreting information than something truly drastic. Peace of mind is something of a given, though I imagine it has to be acquired. I for one wouldn't trust Google Chauffeur the very first time I got behind the wheel.

What Are The Benefits of Self-Driving Cars?

In its current form, where it acts as a supremely helpful form of assisted cruise control, Google Chauffeur simply serves to make driving less stressful and intensive. Levandowski reports that, when he arrives at work, he's ready. Having not just undergone the rigmarole of his daily commute, he has more energy for his job. He'll get more done and probably live longer, too.

Self-driving cars could also be a godsend for those who still have a need for a car yet who might be less adept at driving, such as the elderly or the infirm. This technology could help people to retain their independence and mobility long into life without running the risk of endangering other drivers.

In the long term, when the self-driving cars outnumber the human-driven cars, you're looking for a world in which congestion and traffic lights are a thing of the past. Also, because computers can be programmed to drive with maximum efficiency, CO2 emissions and fuel consumption will also drop.

If you're at all interested in the development and potential of self-driving cars, I heartily advise you to read the Popular Science article. It also contains a lot of information concerning the legal implications of this technology.

Image via Flickr user: Steve Jurvetson