The Good Kind of Problems to Have

Machines will not kill you out of neglect

IEEE Spectrum, the magazine of IEEE, the Institute of Electrical and Electronics Engineers, the self-professed “world’s largest professional organization devoted to engineering and applied sciences,” has a real hard-on for self-driving motor vehicle technology. Five hours ago they ran a story called ‘The 2578 Problems with Self-Driving Cars,’ detailing all the reported cases of autonomous testing vehicles requiring human intervention. If you read that piece, as well as this piece on self-driving cars and bicycles, you will read in it the snark and high standards of the engineering community. They are not against self-driving cars per se, they are simply hammering hard at the limitations of the technology in an effort to set it free or bury it as not practical. That is what engineers do.

Yesterday they ran that second story on challenges the underlying navigational technologies face in identifying and handling bicycles and cyclists on the road. It is a good read and gives a good live-look into the state of the technology as it relates to those of us for whom the bike is a principal mode of transport.

And here is the bottom line: Even in the current imperfect state of bicycle-detection in autonomous vehicle navigation technology, rolling out this technology now would save lives and reduce cyclist injuries.


To summarize the piece, autonomous navigation technologies (ANT’s) are getting really good at identifying vehicles, but not as good at spotting pedestrians and bicycles. This is because pedestrians and cyclists are not uniform in appearance or movement compared to cars, and can act unpredictably when compared to cars.

Of course these are principal complaints of the motor vehicle driver community as well. Why this is the case is something reasonable people can debate. Is it because drivers do not respect the rule of law when it comes to cyclists? Or is it the cyclists that do not? I digress.

While it not clear in the piece, I have surmised that the risk to cyclists vis a vis ANT’s will be directly proportional to the cyclist’s behavior in traffic. In other words, take the lane and be visible and uncertainty in treatment by an ANT decreases, while riding the gutter and darting through moving traffic increases that uncertainty. I will always gamble on the ability of an ANT to see what is directly in front of it better than an object coming in from a side vector.

As those familiar with my philosophy know, the world I strive for is one where all road users observe the rule of law, and I am not disrespected or put at risk for exerting my right to the lane on what usually happens to be a slower, ‘different’ conveyance from the road majority or motor vehicles.

If we are all sharing the same lane, then we do not have to worry about a driver or an ANT hitting me in a right hook from the main lane across the bike lane. This is another way of reiterating the idea that creating parallel infrastructure on the roads creates navigation challenges that neither our human nor machine overlord driver education programs take into account.

So if we could all just share the lane, that would be great.

Additional reference: SF Bikes piece on a ride-along with Uber self-driving cars and how they habitually right-hook the bike lane.

About Ben Folsom

Ben Folsom is a founding member of the Bike Commuter Cabal, a worldwide group of transport cyclists dedicated to protecting the rights of all road users and to encouraging people everywhere to ride more. Ben and his bikes live with his family in Alexandria, Virginia.