London's black cabs are among the most famous in the world. So are their drivers: to get their licence, they have to know the city by heart, without GPS, street by street.
In the coming weeks, a new kind of taxi will start learning London's streets. Without a driver.
What has been announced
On Thursday, Uber and the Chinese company Pony.ai announced that they would be testing Pony.ai's robotaxis in London in the coming weeks. These are seventh-generation cars, capable of driving themselves, initially with a person on board to supervise.
Customers will eventually be able to order them from the Uber app, like any other ride. The two companies are aiming for more than 2,000 robotaxis in Europe.
Not the first in London
Pony.ai will not be alone. Uber is already working in London with Wayve, a young British autonomous driving company. Earlier this year, Wayve launched the UK's first autonomous Uber rides, with around fifteen cars and a driver ready to take over, under licence from London's transport authorities. Uber has even promised to invest a further $300 million in Wayve if its robotaxis are deployed, starting with London.
This is Uber's strategy: not to build its own autonomous cars, but to have several suppliers working on its platform. The company already works with more than 30 companies in the sector, and is aiming for driverless rides in around fifteen cities worldwide by the end of the year.
Europe, a new playground
London is not Pony.ai's first European city. In Zagreb, Croatia, a local company founded by Mate Rimac, known for his electric sports cars, has been running robotaxis equipped with Pony.ai technology since the spring, with a safety operator on board. They can be booked on Uber since August. Pony.ai is also testing its cars in Luxembourg, with Bolt and Stellantis.
In China, the company already runs robotaxis in four major cities. France, for its part, is not yet on the list.
In London, becoming a black cab driver requires passing one of the hardest exams in the world, nicknamed "The Knowledge". Candidates have to memorise around 25,000 streets around the city centre, and thousands of places, so they can plot the best route in their heads. It often takes them three to four years of preparation. Scientists have even observed that this learning changes an area of the brain linked to navigation. An autonomous car, by contrast, learns the city through its cameras, its sensors and highly detailed maps.
The questions that remain
Safety. Robotaxis are making progress, but they are not infallible. In the United States, Tesla's steering-wheel-free cars have already been the subject of a federal investigation barely after launch. London, with its narrow streets, its buses, its cyclists and its left-hand driving, will be a demanding test.
The data. A self-driving car continuously films everything around it. Seeing Chinese technology map the streets of a European capital will inevitably raise the question of where those images go, and who has access to them.
Jobs. Taxi and private-hire drivers are among those most directly affected by the arrival of self-driving cars. The transition will be slow, with operators on board at first, but the direction is clear.
What we take away
In 2017, researchers predicted that an AI would be able to drive a lorry by 2027. When we checked their predictions, the answer was: "partly". The arrival of robotaxis in London shows that this part is growing, city after city.
The question is no longer really whether driverless cars will arrive in Europe. It is at what pace, under what rules, and with whose technology.
>>