What will self-driving cars mean for cyclists?

A cyclist passes a Google self-driving car at Mountain View, California, back in 2012. Image: Getty.

Last week, I joined thousands of other Brits in hopping on my bike to make the most of the uncharacteristically warm weather. But just as I was remembering all of the things I love about cycling, I was rudely reminded of one of its major problems.

It’s a scene that doesn’t need much setting because it happens far too often. I was pedaling down a typical London street, one lane of traffic moving in each direction. An engine revs behind me – an impatient driver looking to fill the two car-lengths between my bike and the vehicle in front. Overtaking will do no good here, and besides, there are cars coming in the opposite direction. It would be an unsafe maneuver.

The revving gets louder, and suddenly I feel the car whisk past my shoulder with millimetres to spare, squeezing between me and the oncoming traffic. It’s so close I’m destabilised and narrowly avoid a crash. All too aware of London cyclists’ bad reputation for shouting profanities at drivers, I keep my anger to myself. But an unexpected thought springs to mind: I can’t wait for self-driving cars.

My reaction was perhaps well-founded. In 2016, 102 cyclists were killed and a further 3,397 seriously injured on Britain’s roads. Whilst riding a bike remains safe by statistical standards – with only one death per 30m miles cycled on Britain’s roads, and the general health benefits far outweighing the relative risk – every cyclist has a story of a hairy experience.

Proponents of self-driving cars promise they will reduce that epidemic to close to nil. Through the combined functions of automatic braking, hazard detection, avoidance of driver fatigue and the elimination of blind spots, the technology does seem promising.

However a recent spate of deaths in the U.S. casts doubt on my rosy assumption that autonomous vehicles will solve cyclists’ problems once and for all. On the night of 18 March, an Uber self-driving car struck and killed a woman wheeling a bicycle across a road in Arizona. Five days later, a Tesla car on autopilot mode crashed in California, killing its driver.

It is clear that autonomy, in its current form, is far from perfect. Vehicles’ detection systems are developing fast but are still primitive, and in cases where cars offer partial autonomy in the form of steering assists and cruise control, the risk is that drivers can lose concentration. What’s more, when autonomous vehicles have to operate on the same roads as unpredictable road users – like cyclists and pedestrians – they face a far trickier job.


Though autonomous cars may be on Britain’s roads as early as 2019, it will be many years before every vehicle is automated. “The transition is going to be really messy,” Roger Geffen, the policy director of the advocacy group Cycling UK, tells me. “Before autonomous cars can really share the streets with pedestrians and cyclists, they’ve got to not just detect their presence but predict their movements. Cyclists negotiate space with drivers by a combination of eye contact and hand signals. How are driverless cars going to understand that?”

Until technologists can find an answer to that question, Geffen’s fear is that pedestrians and cyclists will be demonised for their unpredictability, possibly even facing the prospect of being banned from certain roads. And even if technologists could design an algorithm that can detect cyclists and pedestrians in every instance, autonomous vehicles still raise unanswered questions about cyclists’ place on the roads.

Looking to the future, there are two possible extremes. One is utopian: the lack of need for a driver will mean a small fleet of driverless cars working around the clock could replace the thousands of cars lying idle on our streets, freeing up space for cycle infrastructure and pavements.

But that scenario is not inevitable. “The nightmare future,” Geffen explains, “is one where the manufacturers are determined to recoup their investment by trying to make sure everybody’s got a self-driving car. We’ll end up with complete gridlock and the technology never getting to the point where it’s able to detect the presence of pedestrians and cyclists.”

Driverless cars offer great promises, and it seems fair to assume they will eventually lead to a reduction in road fatalities. But it would be foolish to expect that to come soon, and we may see an increase before numbers start to fall. It is likely cyclists and pedestrians will have to fight for their right to remain unpredictable, and possibly learn new behaviours to interact with self-driving vehicles.

One thing, however, is certain. The roads are going to change.

 
 
 
 

Podcast: Beyond the wall, with John Lanchester

A sea wall in Japan. Image: Getty.

This week it’s another live episode, of sorts. In early April I was lucky enough to chair an event at the Cambridge Literary Festival with the journalist and novelist John Lanchester.

John was mostly there to promote his latest novel, The Wall, a “cli-fi” book about a Britain trundling on after catastrophic climate change has wiped out much of the planet. In the past he’s also written about other vaguely CityMetric-y topics like the housing crisis and the tube - so he’s a guest I’ve been hoping to get on for a while, and was kind enough to allow us to record our chat for posterity and podcasting purposes.

Incidentally, I didn’t find a way of turning the conversation to the tube. We do lose ten minutes to talking about Game of Thrones, though.

The episode itself is below. You can subscribe to the podcast on AcastiTunes, or RSS. Enjoy.

Jonn Elledge is the editor of CityMetric. He is on Twitter as @jonnelledge and on Facebook as JonnElledgeWrites.

Want more of this stuff? Follow CityMetric on Twitter or Facebook.