Should Cyclists Fear the Rise of the Robot Car?

Rapha x Killik & Co

Guest content by Rapha for Killik & Co. Words by Laura Laker.

What’s all the fuss about ‘autonomous vehicles’? According to Michael Bloomberg, the former mayor of New York, driverless, or autonomous, vehicles (AVs) could be one of the most exciting developments ever to happen to cities.

An AV is a car, van, bus or truck capable of navigating and responding to its environment with limited or no input from any human occupants. AVs are graded on a 0-5 scale of autonomy: Level 0 is a 20-year-old Ford Fiesta; 3 still requires a driver but takes command of all safety-critical functions in certain traffic conditions; 5 is fully autonomous, with no driver at the wheel (because there is no wheel).
Traditional car manufacturers and software giants alike are racing to build the first fully autonomous model to successfully, and legally, take to the roads. According to its cheerleaders, the AV will bring many benefits. It’s likely there will be many fewer road deaths – according to the World Health Organisation, flesh-and-blood drivers cause about 1.25m deaths each year, whereas driverless cars are programmed to defer to vulnerable road users such as pedestrians and cyclists. Without unpredictable, error-prone humans in control, the idea is that AVs will communicate with each other to form fast-moving, closeformation, impeccably behaved convoys, which will take up less space than human-piloted cars, vans and trucks.

As a result, the theory goes, there will be better traffic flow and less need to expand major road networks. And if AV ride-sharing is adopted en masse, it is thought that many of us will give up our private cars, further reducing vehicle traffic.

Elon Musk, the co-founder and CEO of the trailblazing car manufacturer Tesla Inc., has said that one day AVs will be as common as elevators, and as safe. “It’ll be an order of magnitude safer than a person [driving a car],” he told a technology conference in 2015.
“We await the decisions of governments around the world, who are still assessing the viability of the AV”
That all sounds better for cyclists – is it? That depends. There is an argument that road safety could be dramatically improved if some of the technologies already developed by AV companies, such as autonomous emergency braking, were routinely fitted to all new traditional cars. “Vision zero [the target of zero road deaths] could genuinely be within reach if you put some of these technologies in [traditional] motor vehicles,” says Ceri Woolsgrove of the European Cyclists’ Federation.

Failing that, we await the decisions of governments around the world, who are still assessing the viability of the AV on the open road. AV companies are looking to unleash Level 4, highly autonomous vehicles, by the end of next year, with full Level 5 automation by 2020 or 2021. That seems optimistic; Level 3 AVs require a human driver to intervene at the controls in trickier scenarios, but there is evidence that humans become over-reliant on the technology and fail to respond appropriately when needed, i.e steering the vehicle through a complex scenario involving cyclists and pedestrians.
What about reducing traffic on the roads? A small study early this year in Denver, Colorado, suggested that ride hailing apps encourage users to forego walking, cycling and public transport in favour of a cab; another in New York concluded that ride-hailing app drivers in the city are increasing traffic unsustainably. By contrast, an MIT study this year suggested that the ride-sharing capabilities of Uber, Lyft and others, properly used, could significantly reduce the number of cabs required, and that the advent of AVs could reduce their numbers further still.
“One of the most successful obstacle detection systems can spot a cyclist only 74% of the time”
Are bicycles a particular problem for autonomous vehicles? Yes. Cyclists present AVs with a challenge: riders can change speed and direction rapidly, and come in a variety of shapes and sizes, none of them simple for AVs to anticipate and detect. One of the most successful obstacle detection, systems, Deep3DBox, developed by Cornell University, can spot a cyclist only 74% of the time in test conditions, and correctly deciphers which way they are facing just 59% of the time.

To put the task of the AV developers in context, it’s thought a fully autonomous vehicle will require 50 times the processing power used to process the data generated by advanced driver assistance systems, such as lane assistance, fitted in many current traditional cars. So let’s put it another way: AVs are a problem for cyclists.
“Google is one of the major players in the AV race, and Waymo is its AV arm”
What are autonomous vehicle companies doing about cyclists? None of the AV companies I contacted wanted to discuss cyclists specifically. Google is one of the major players in the AV race, and Waymo is its AV arm. Waymo’s AV testing programme was forced back to the drawing board when one of its cars encountered a cyclist doing a track stand on the road; the car froze (programmed to defer to vulnerable road users, it didn’t know how to proceed).

The company set to work on everything from detecting unicyclists to deciphering cyclists’ hand signals. It has even mocked up a cityscape 100 miles east of California’s Silicon Valley to test its vehicles. Waymo is also considering cyclist tagging, either via smartphone apps or electronic devices. But this supposes every cyclist has a fully charged phone or that the bike’s electronic tag is working.
When can we expect to share the roads with autonomous vehicles? Soon. Bosch, the German tech company, predicts there will be 150m vehicles with some level of autonomy on the world’s roads by 2025. (As of 2010 there were more than 1bn vehicles on the planet’s roads.) In the UK, testing of driverless trucks will begin next year.

Musk claims a fully autonomous Tesla will be ready in 2018 but that regulatory approval may take up to a further three years. Government legislation around the world currently holds back the march of AVs. The public are uncertain too: in a recent survey by the UK’s Institution of Mechanical Engineers 66% of drivers said they would be uncomfortable travelling at 70mph in a driverless vehicle.
What will future city roads look like? Pedestrians will be able to bring AVs to a halt by simply stepping out in front of them and triggering their automatic braking function. As a result academics and experts I spoke to from both sides of the debate saw designated AV only expressways as highly probable. Were such segregation to be implemented in cities it could be bad news for cyclists. As Woolsgrove puts it: “[Cyclists] want to use those nice direct routes as well.”

Last summer Musk tweeted he had “verbal government approval” for a superfast underground ‘hyperloop’ between New York, Philadelphia, Baltimore and Washington, DC; it turned out to be hot air but it illustrated the ambition in this area. Jaworzno in Poland, for instance, is considering turning a former railroad track into a hyperloop test track as part of plans to make the entire city an autonomous car zone. In this scenario there is a risk that legislators will yield to the temptation to rule obstruction of AVs, whether by cyclists or pedestrians, an offence.
“The appeal of promoting widespread urban cycling is evident: public health is enhanced, congestion is reduced, air quality is improved”
Is a compromise between autonomous vehicles and other road users likely? AV companies have a lot to gain from worldwide adoption of this technology. What’s more, governments around the world will be competing to attract investment from the big AV firms.

But traffic does not create great cities. As Woolsgrove puts it: “Public authorities are looking to shift from private motorised transport to walking, to cycling, to public transport, whereas driverless car manufacturers are looking to provide easier access to on-demand private transport.”

The appeal of promoting widespread urban cycling is evident: public health is enhanced, congestion is reduced, air quality is improved. According to Robin Hickman, reader in transport and city planning at University College London: “The case is overwhelming for increasing cycle-mode share to 30% or more in urban areas. That takes 20 or 30 years of consistent investment.”

It might be a while before the bicycle and the autonomous vehicle happily share the same road. Watch this space.
This article is a guest piece written by Laura Laker for Rapha and first appeared in issue 006 of Mondial. Rapha also joined us during House of Killik Soho for a talk on The Experience Economy as well as hosting a Winter Bike Maintenance Workshop.

To be the first to hear about future events please sign up to our newsletter below.