Science

Waymo nearly hit a cyclist, raising fears

Waymo nearly – Waymo driverless cars are operating in London without passengers, with a human ready to take control. After a near miss at a roundabout, the company said its system detected a cyclist and a safety driver took manual control to brake. London Cycling Campaign wa

For months, Waymo’s driverless cars have been moving through London’s streets without taking passengers—always with a human in the loop, ready to seize control if the automated system needs help. The machine, in day-to-day encounters, has struck one cyclist as cautious and predictable.

Then came a near miss.

Cycling home. the cyclist was circling a roundabout when a Waymo vehicle began to pull onto it in front of them. The rider says they had priority. They also say the car didn’t appear to be slowing—at first it looked like it was heading straight for them. In a moment that split the ride into before and after, the vehicle suddenly stopped.

The cyclist stressed the incident wasn’t the worst they’d experienced that day. and that they didn’t feel in “serious danger.” They had seen what was happening and were prepared. What stayed with them wasn’t just the risk. but the fact that the decision came from an automated system rather than a person.

Out of curiosity, they contacted Waymo with the time and place and asked what happened. A company spokesperson explained that the automated driving system detected “a cyclist travelling near the outer edge of the roundabout approaching from the right.” The safety driver. the spokesperson said. assumed manual control “to apply the brakes at the entrance of the roundabout.” In the company’s telling. the AI didn’t look like it was going to stop. so a human jumped in.

The cyclist then asked how often that kind of intervention happens in London. Waymo did not provide a figure. saying it doesn’t have that data and also doesn’t want to focus on it as a metric. They also asked what confused the system—particularly given that they say they were cycling legally and that their bike is neon yellow—and what would be done to improve the algorithm based on the incident. They said they received no response.

Other cyclists in London described more positive encounters. Several said the Waymos have acted well around them. One rider, however, recalled an incident near roadworks where a sign reading “NARROW LANES DON’T OVERTAKE CYCLISTS” appeared to prompt a Waymo to do exactly that.

Simon Munk of the London Cycling Campaign said that, overall, the cars have been acting cautiously around cyclists. Still, he sees potential trade-offs that can’t be ignored. Driverless systems could make streets safer and more predictable. he said. but they could also become a menace that pushes vulnerable road users further to the margins.

Munk points to what he called the “different level” of London. Waymo’s testing and public demonstrations have often highlighted situations designed to stress the technology—he recalled the company showing him a video of a car successfully navigating a junction the company described as chaotic in San Francisco. But he says his own commute in London has included at least half a dozen worse moments than the one Waymo used to showcase its capabilities.

He’s also uneasy about what happens after launch. Even if driverless cars seem safe at first, Munk worries that later software updates could change their behaviour. In his view, driverless cars can’t simply be signed off as safe. They will need close and constant scrutiny to keep companies honest and protect citizens.

The cyclist who contacted Waymo doesn’t claim the incident has changed their mind about the broader debate. They said they wrote about AI drivers last year, knowing tech companies were bringing cars to London. Their earlier argument was that they may be better off surrounded by unpredictable and imperfect AI drivers than by tired. distracted. and angry human drivers. They’ve been knocked off their bike several times by people. The near miss with Waymo, they say, didn’t overturn that view. Still. they note that they once saw a person “sail through police tape” blocking a road. and the company said a person was in control at the time.

In the cyclist’s assessment. they don’t fully trust AI to operate near them. but they also don’t fully trust people. Their faith, they said, is that AI can keep learning relatively quickly. Sensors that help these models spot hazards, they added, should get more accurate and powerful. AI may or may not be the lesser of two evils now. but they believe it can become the safer option in the future.

London, though, isn’t just waiting for Waymo. The city is preparing for more driverless players. Wayve driverless taxis could arrive within the next few months, and whether they take a cautious approach remains an open question.

Munk frames it as a competition and a calibration problem. Just as one driver can be safer than another, one AI can be safer than another. And just as a human driver can get tired and careless. he says. AI can be tweaked at will to act however a company wants. He pointed to an example from last year. when Waymo made its cars more assertive—an update intended to stop the vehicles slowing traffic. In practice. Munk said. the result was behaviour compared to “an aggressive New York taxi driver. ” and he said it made many San Franciscans wary.

There are also what Munk sees as troubling ideas about shifting responsibility onto the people outside the vehicle. He said they’ve already seen “solutions” to road safety that offload the problem to pedestrians—suggesting they all wear special vests to make life easier for autonomous cars.

For Munk. the central fear is not just what driverless cars do today. but what companies may choose to do tomorrow once these systems become routine. He posed the question of whether companies. pressed by performance goals and market pressure. will decide to speed up journeys. cut corners. and act more aggressively—after all. he said. rivals could land with promises of faster and cheaper services by being more assertive.

“Five years from now [or] 10 years from now. when these things are just everyday. do [these companies] go ‘do you know what. our passengers are getting slow journeys. do we just speed them up a little bit [and] cut a few corners?’” Munk asked. “Or does a rival service land and say ‘we’re 20 per cent faster than Waymos and 20 per cent cheaper’. and they do that by speeding up 20 per cent and being 20 per cent more aggressive?”.

His conclusion was blunt: the reality, he said, is that vigilance won’t be optional.

Waymo driverless cars London cycling safety automated driving system roundabout AI road safety London Cycling Campaign Simon Munk Wayve software updates

4 Comments

  1. So it “nearly hit” but then stopped… cool cool. I feel like these companies always say the system detected something like that makes it fine. If it didn’t slow at first, how is that “detected”? Also London traffic is already chaos.

  2. I don’t get why they’re even driving around with no passengers if they’re that worried about cyclists. Like isn’t that the whole point of testing? Maybe the safety driver was late? Or maybe the cyclist cut in and blamed the car… hard to say from a headline.

  3. “Human ready to take control” is exactly what everyone says right before something goes wrong. A roundabout is where you NEED to be predictable, and if it looked like it was going straight at them then yeah, that’s scary. Glad they didn’t feel seriously in danger, but “didn’t appear to be slowing” is still not great. Next thing you know people will be acting like this is normal because it didn’t actually hit them.

Leave a Reply

Your email address will not be published. Required fields are marked *

Are you human? Please solve:Captcha


Secret Link