Will robotaxis clog up Britain’s cities?

For almost a decade, London mayor Sir Sadiq Khan has sought to reduce car use — expanding the city’s congestion charge, promoting cycling and limiting rises in bus fares.

But the capital has remained one of Europe’s most congested cities, with drivers spending 91 hours in traffic last year, according to analysts Inrix.

Now Transport for London, the city’s transport authority, has identified a new challenge: the introduction of autonomous vehicles.

Earlier this year, it raised concerns about how robotaxis would fit in with Khan’s aim for 80 per cent of journeys to be made by cycling, walking or public transport by 2041.

“It is really important that we get [the rollout of robotaxis] right so that we can see it complement our ability to reach that target rather than undermine it,” Christina Calderato, TfL’s director of strategy, told the London Assembly’s transport committee in July.

Robotaxi operators, which are hoping to start services in London by early 2027, have emphasised their small scale. Waymo, owned by Alphabet, parent company of Google, said its test fleet was below 100 vehicles and its approach was to “grow gradually”.

Rise of the robotaxis

This is the second part of a series on how London’s streets will see a rise in the number of autonomous vehicles in the coming months.
Part I: London’s cabbies prepare to battle the robotaxis
Part III: Waymo, Wayve or Baidu: whose robotaxis will win out?

“If we were given an operating licence, I suspect that we would be operating in the low hundreds of vehicles compared to the 120,000 taxi and private hire vehicle drivers that there are in London,” Waymo’s head of policy for UK and Europe Ben Loewenstein told the London Assembly in June.

Wayve, a British start-up backed by Uber, also said it would start with “a small number of vehicles” to ensure any increase in congestion would be minimal.

“As the service grows, smarter routing, more efficient driving and more people choosing shared autonomous rides will help reduce overall congestion,” it added.

Siddartha Khastgir, head of safe autonomy at the University of Warwick, estimated that robotaxi numbers could reach between 500 and 1,000 within five years of their initial rollout.

“It’s not going to increase congestion,” Khastgir said. “The only places that I can think of congestion increasing are when an incident happens and people don’t know how to react to it but that would be very sporadic and ad hoc.”

Such incidents have already occurred in the US where Waymo operates more than 3,800 vehicles across 10 cities.

Last December, for example, a power outage disabled traffic lights in San Francisco, causing confused Waymos to stall at intersections. Videos posted on social media showed drivers trying to go around the stopped taxis, but this would be difficult in central London where many roads are less than six metres wide.

Khastgir said the UK’s Home Office and the Department for Transport were working to train emergency services to respond to similar incidents. All of the robotaxis to be deployed in London will have remote support from UK-based humans to respond when the vehicles malfunction or get stuck.

“In those extreme events these vehicles are not performing as well as . . . human-driven vehicles,” said Matthew Raifman, research lead in policy and autonomous vehicles at the University of California, Berkeley.

Raifman suggested that cities could issue a geofence — a virtual boundary — to restrict robotaxis from entering certain areas during large events such as concerts. He added that communication between city authorities and operators would also be key to preventing robotaxis worsening congestion.

Some experts have also warned of congestion increasing due to autonomous vehicles driving around empty.

A study published in the journal Transport Findings in May analysed about 14mn Waymo trips in California, equivalent to 86mn vehicle miles travelled, using data that the company reported between August 2023 and December 2025.

Approximately 46 per cent of the miles were “deadhead” miles, where the vehicle travelled without a passenger, the study found. This is a similar rate of “empty miles” that was reported by ride-hailing operators such as Uber and Lyft.

Awad Abdelhalim, author of the study and associate director at Massachusetts Institute of Technology Transit Lab, said the findings showed that — like human-driven taxis — Waymos were on the road more than the average private car.

However, he added, Waymo fleets remained far too small to be a meaningful contributor to congestion in the UK. But as robotaxi fleets scale up, this could become an issue that needs to be addressed, said Abdelhalim.

Waymo acknowledged that “empty miles” could not be eliminated entirely but added that it had “a direct commercial incentive” to minimise them since it shoulders the cost of every empty mile driven, which leads to higher operating costs and energy consumption.

Researchers say one way to reduce ‘deadheading’ is for cities to allow more autonomous vehicles to park at kerbside spots instead of driving around empty waiting for passenger pick-ups. This would also minimise congestion, they say.

While lessons from robotaxi services in US cities are likely to be reflected in the UK deployment, some experts caution that conditions in London are starkly different due to streets that are narrower and not part of a grid system.

“There are lots of differences in London and there needs to be some real-world understanding of what this looks like when the autonomous vehicles are on the road,” a person close to TfL said.