Driverless private hire by 2021
Driverless private hire vehicles could begin carrying passengers around London within just three years under ambitious plans announced by Addison Lee and British autonomous-driving specialist Oxbotica. The companies intend to map hundreds of thousands of miles of roads and develop self-driving services that could eventually include airport transfers, shared journeys and corporate shuttle services.
London's private hire industry could be approaching one of the biggest technological changes in its history.
Addison Lee has announced a partnership with autonomous-vehicle technology company Oxbotica aimed at bringing self-driving passenger services to the capital by 2021.
If successful, the project could eventually allow passengers to book a vehicle that arrives without a conventional driver behind the wheel.
But before autonomous vehicles can begin navigating one of the world's most complicated cities, the companies have an enormous amount of work to do.
And it begins with a map.
The first stage of the project involves creating highly detailed digital maps of London's road network.
Addison Lee and Oxbotica intend to map more than 250,000 miles of public roads in and around the capital.
These aren't ordinary navigation maps.
The system needs to understand the physical environment in considerably greater detail than a human using satellite navigation.
Information will include the precise location of kerbs, road signs, traffic lights, road markings, junctions, and other permanent landmarks.
The objective is to create a digital representation of London detailed enough for autonomous-driving software to understand exactly where a vehicle is located.
A human driver looks through the windscreen and interprets the road almost instinctively.
They recognise a traffic light.
They see the kerb.
They understand where one lane ends and another begins.
They notice a pedestrian approaching a crossing.
For a computer, all of that information has to be detected, classified and interpreted.
Autonomous vehicles therefore combine detailed mapping with sensors and sophisticated software.
The vehicle needs to determine its precise position while simultaneously identifying everything happening around it.
London makes that particularly difficult.
Developing an autonomous vehicle on a wide American highway is one thing.
Sending one through central London is something else entirely.
London's streets contain almost every challenge imaginable.
Narrow roads.
Complex junctions.
Pedestrians.
Cyclists.
Motorcycles.
Black cabs.
Private hire vehicles.
Buses.
Delivery vans.
Roadworks.
Parked vehicles.
Temporary diversions.
Bus lanes.
And streets designed centuries before the invention of the motor car.
A successful autonomous vehicle has to interpret all of them safely.
For the project, Addison Lee is partnering with one of Britain's best-known autonomous-driving technology companies.
Oxford-based Oxbotica specialises in software allowing vehicles to operate autonomously.
The company's technology is designed to work across different types of vehicle rather than being tied to a single manufacturer.
Oxbotica had already been involved in autonomous-vehicle testing in Britain before the Addison Lee partnership.
The companies believe combining autonomous-driving technology with Addison Lee's existing transport operation could accelerate the move from experimental vehicles to commercial passenger services.
Addison Lee already has vehicles, maintenance facilities, operational experience and customers.
Oxbotica brings the autonomous technology.
The timetable is ambitious.
The companies say they are working towards providing self-driving services in London by 2021.
That doesn't necessarily mean every Addison Lee vehicle will suddenly become driverless in three years.
Initial services are expected to be much more controlled.
Possible early applications include airport transport, corporate shuttles, campus services and shared passenger journeys.
These types of routes could provide a more manageable environment for introducing autonomous technology before attempting unrestricted journeys throughout London.
Airport transportation is particularly interesting.
Airports generate enormous numbers of relatively predictable journeys.
Passengers repeatedly travel between terminals, hotels, business districts, car parks and transport hubs.
Some airport environments also contain controlled roads where traffic is more predictable than on a typical central London street.
This could make airport-related services an attractive early application for autonomous vehicles.
Oxbotica has already been involved in autonomous technology projects connected with both Heathrow and Gatwick.
One possible future service might operate between a Heathrow hotel and an airport terminal.
Instead of calling reception for a shuttle or booking a conventional vehicle, a passenger could request an autonomous vehicle.
The vehicle arrives.
The passenger confirms the booking.
The luggage is loaded.
The vehicle follows a predefined route to the terminal.
No conventional driver would necessarily be required.
A service like this would be considerably easier to introduce than an autonomous vehicle expected to accept any destination anywhere in Greater London.
A full airport transfer into London presents a much greater challenge.
Consider a journey from Heathrow to a hotel in Soho.
The autonomous vehicle would need to navigate airport roads.
Then motorway traffic.
Then London's suburban streets.
Eventually it would enter congested central London.
It might encounter cyclists, buses, pedestrians, roadworks, temporary restrictions and delivery vehicles.
Finally, it needs to stop safely at the correct location for the passenger.
That is considerably more demanding than operating a repetitive shuttle route around an airport.
Business campuses present another obvious opportunity.
A company may need to transport employees repeatedly between a railway station and an office complex.
The route could be relatively short.
Traffic conditions may be predictable.
Passenger demand can be forecast.
And the vehicle repeatedly travels between the same locations.
These characteristics make shuttle services attractive as an early stage in autonomous passenger transport.
Addison Lee also believes autonomous technology could create new forms of shared transport.
Instead of one passenger hiring an entire vehicle, several people travelling in the same direction could share it.
Software would calculate efficient routes and determine where passengers should be collected and dropped off.
Removing the conventional driver could eventually reduce one of the largest costs associated with operating a passenger vehicle.
That could potentially make shared road transport significantly cheaper.
This is one of the industry's biggest questions.
Drivers represent a substantial part of the cost of providing private hire services.
An autonomous vehicle doesn't need wages.
It doesn't require breaks.
And, theoretically, it could operate for much longer periods each day.
Addison Lee chief executive Andy Boland said autonomous vehicles could eventually produce savings through better utilisation of the company's vehicles.
But driverless vehicles will create other costs.
The technology itself is expensive.
Vehicles require sophisticated sensors and computing equipment.
Software has to be maintained.
Vehicles still need cleaning, servicing, tyres, insurance and repairs.
And entirely new operational and regulatory systems may be required.
Driverless doesn't mean cost-free.
For London's private hire drivers, the announcement inevitably raises a difficult question.
What happens to their jobs?
Addison Lee said its approximately 5,000 UK drivers would continue doing what they were doing for the foreseeable future.
The company distinguished its premium, driver-operated services from possible new autonomous products.
In other words, the initial vision isn't necessarily to remove every driver from every Addison Lee vehicle.
Instead, autonomous vehicles could create additional types of transport alongside conventional chauffeur and private hire services.
There are good reasons why professional drivers could remain important.
A premium private hire journey isn't simply about moving a vehicle from one coordinate to another.
Drivers help passengers with luggage.
They meet people at airports.
They identify the correct passenger.
They adjust to unexpected circumstances.
They assist elderly travellers.
They communicate with hotels and reception desks.
And they provide reassurance to visitors unfamiliar with London.
A computer can potentially drive a car.
Replacing every aspect of a professional driver's role is a much more complicated task.
Consider an international passenger arriving at Heathrow.
The aircraft lands late.
Immigration takes 45 minutes.
The passenger's suitcase doesn't appear.
Eventually they enter the arrivals hall and look for their driver.
A traditional Meet & Greet service can adapt.
The driver can monitor the flight.
Wait in the terminal.
Help identify the passenger.
Assist with luggage.
And guide them to the vehicle.
An autonomous car sitting somewhere in a car park cannot easily reproduce all of those human interactions.
This suggests airport transfers could eventually divide into different categories.
Simple journeys could become increasingly automated.
Premium Meet & Greet services could continue using professional drivers.
The announcement also needs to be understood in the context of London's changing private hire industry.
Smartphone booking has transformed the market.
Uber has expanded rapidly.
Traditional minicab operators have introduced apps and online booking.
Passengers increasingly expect fast booking, live vehicle information and electronic payment.
Autonomous vehicles could represent the next stage of that competition.
If the cost of providing a journey can eventually be reduced by removing the driver, the economics of private hire could change dramatically.
Addison Lee isn't alone.
Uber has invested heavily in autonomous vehicles.
Toyota recently announced a $500 million investment in a partnership intended to develop autonomous vehicles for Uber.
But the development of self-driving technology has also demonstrated the potential dangers.
In March 2018, an autonomous Uber test vehicle struck and killed a pedestrian in Arizona.
The incident attracted worldwide attention and raised difficult questions about the safety of testing autonomous vehicles on public roads.
Any company attempting to introduce driverless passenger services in London will therefore face intense scrutiny.
Passengers will not adopt autonomous transport simply because the technology is impressive.
They need to trust it.
A driverless vehicle must be capable of dealing with situations that human drivers encounter every day.
A child runs towards the road.
A cyclist moves unexpectedly.
A delivery van blocks a lane.
A traffic light fails.
A police officer manually directs traffic.
Roadworks alter the normal route.
Heavy rain reduces visibility.
Someone opens a car door.
These events are ordinary parts of urban driving.
An autonomous system needs to recognise them and respond appropriately.
Anyone who drives professionally in central London knows how unpredictable pedestrians can be.
People cross between stationary vehicles.
Tourists look the wrong way.
Commuters rush across roads.
Passengers step into cycle lanes while looking at their phones.
A human driver learns to anticipate behaviour that isn't always technically correct.
Autonomous vehicles need to develop similar defensive capabilities.
That could prove one of the most difficult parts of operating safely in London.
London's black cab drivers are famous for their knowledge of the capital.
Drivers undertaking the Knowledge learn thousands of streets and points of interest.
An autonomous vehicle approaches the same problem very differently.
Instead of memorising London in a human brain, the vehicle uses digital maps, sensors, localisation technology and software.
Both are attempting to answer the same fundamental questions: where am I, where is my passenger going, and what is the safest and most efficient way to get there?
The methods could hardly be more different.
There is also an important terminology issue.
Headlines may describe Addison Lee's planned vehicles as “driverless taxis”, but Addison Lee operates in the private hire market.
A London taxi — normally a black cab — can be hailed in the street or taken from an authorised taxi rank.
A private hire journey must be booked through a licensed private hire operator.
Autonomous technology doesn't automatically erase that distinction.
If a driverless vehicle operates as private hire, passengers would still need an appropriate booking arrangement through the operator under the regulatory system applicable at the time.
This may become one of the most difficult regulatory questions.
Today, responsibility is relatively easy to understand.
There is a driver.
There is a licensed vehicle.
There is an operator.
But what happens when there is no driver?
If an autonomous vehicle makes a mistake, who is responsible?
The passenger?
The operator?
The vehicle manufacturer?
The company that developed the software?
The company that supplied a sensor?
These questions will need clear answers before genuinely driverless commercial vehicles can become commonplace.
Motor insurance has traditionally been built around human drivers.
Autonomous vehicles change that assumption.
An accident might be caused by software rather than human error.
Sensor failure could become as important as driver negligence.
Cybersecurity may become a motor-insurance issue.
Software updates could affect vehicle behaviour.
Insurers will therefore need to develop new ways of assessing risk.
The autonomous-vehicle revolution isn't only a challenge for car manufacturers and private hire operators.
It could transform the insurance industry as well.
A conventional car can be stolen.
A connected autonomous vehicle introduces another possibility: it could potentially be attacked digitally.
If vehicles rely heavily on software, connectivity and digital maps, cybersecurity becomes a fundamental safety requirement.
Manufacturers and operators will need to protect vehicles from unauthorised access.
The more control computers have over steering, acceleration and braking, the more important those protections become.
Supporters argue that self-driving technology could make roads more efficient.
Vehicles could communicate with transport systems.
Routes could be optimised.
Ride-sharing could mean fewer cars carrying only one passenger.
Vehicles could potentially spend less time searching for parking.
But there is another possibility.
If autonomous journeys become cheap and convenient, people might take more car journeys.
Someone who currently walks, cycles or uses public transport might instead order a driverless vehicle.
Empty autonomous vehicles could also travel between bookings.
That could actually increase traffic.
Technology alone therefore doesn't guarantee less congestion.
The environmental impact will depend partly on the vehicles used.
If autonomous fleets are electric, they could help reduce local exhaust emissions.
If ride-sharing increases vehicle occupancy, fewer cars might be needed to carry the same number of passengers.
But if driverless transport creates millions of additional journeys, some of those advantages could disappear.
London's transport authorities will need to consider not simply whether autonomous vehicles work, but how they fit into the wider transport network.
Even the most optimistic autonomous-vehicle scenario doesn't eliminate the need for mass public transport.
One Underground train can carry hundreds of passengers.
A railway can move enormous numbers of people into central London.
A bus uses road space much more efficiently than dozens of individual cars.
Autonomous private hire therefore seems more likely to complement London's transport network than replace trains, Underground services and buses.
Its greatest opportunity may be filling gaps between them.
Transport planners often talk about the first mile and last mile.
A passenger may have an excellent railway connection for most of their journey.
The difficult part is getting from their home to the station or from the station to their final destination.
Autonomous shared vehicles could potentially solve that problem.
A passenger might take a driverless shuttle to a railway station.
Travel across London by train.
Then use another autonomous vehicle for the final part of the journey.
That could make public transport more attractive rather than competing directly with it.
Of all the possible applications, airports remain particularly interesting.
They combine enormous passenger demand with predictable destinations.
Heathrow has terminals, hotels, railway stations, car parks and businesses concentrated within a relatively small area.
Gatwick has similar characteristics.
Driverless shuttles operating between these fixed locations could provide valuable experience before autonomous vehicles tackle completely unrestricted journeys.
And once the technology proves itself in controlled environments, the operating area could gradually expand.
Eventually, perhaps a driverless car could travel between Heathrow and Gatwick.
But that would be considerably more complicated than an airport shuttle.
A Heathrow-to-Gatwick transfer requires motorway driving, junctions, variable traffic conditions and navigation through two separate airport road systems.
An autonomous vehicle would also need to deal with incidents, diversions and changing weather.
The technology may ultimately become capable of doing all of this.
The question is when.
The planned mapping of more than 250,000 miles of roads sounds enormous.
But collecting the data is only the first step.
The maps need to remain current.
London changes constantly.
Roadworks appear.
Junctions are redesigned.
New cycle lanes open.
Road restrictions change.
Traffic lights move.
Buildings are demolished and constructed.
An autonomous mapping system therefore cannot simply photograph London once and consider the job finished.
It needs to understand a city that continually changes.
There is a significant commercial prize for companies that solve these problems.
The UK connected and autonomous vehicle market has been forecast to be worth approximately £28 billion by 2035.
That explains why established transport companies, technology businesses and vehicle manufacturers are investing heavily.
Addison Lee doesn't want to watch that transformation from the sidelines.
Its partnership with Oxbotica is an attempt to ensure that a traditional ground-transport operator has a role in whatever comes next.
Successfully operating autonomous passenger vehicles in London would be significant internationally.
A system capable of handling London's streets could potentially be adapted for other complex cities.
Oxbotica chief executive Graeme Smith described the partnership as a major step towards bringing autonomous vehicles into mainstream use in London and eventually elsewhere in Britain and internationally.
The companies are already looking beyond the capital, with New York among the cities identified as a potential future market.
The most striking part of the announcement is the timetable.
2021 is only three years away.
Developing technology that can safely navigate London's streets in that period represents an extraordinary challenge.
Regulation needs to develop.
Insurance needs to adapt.
Vehicles need to be selected.
Maps need to be created.
Software needs to be tested.
Passengers need to trust the technology.
And the economics have to make sense.
The ambition is substantial.
Whether reality can keep pace remains to be seen.
For professional private hire drivers, the announcement will understandably attract attention.
But there is little reason to expect human drivers to disappear from London's roads overnight.
Autonomous technology is likely to arrive gradually.
Controlled shuttle services may come first.
Simple routes may be automated before complicated ones.
Premium and Meet & Greet services may retain drivers much longer.
And passengers themselves may continue to prefer a human driver for certain journeys.
The motor car didn't eliminate every previous form of transport immediately.
Smartphone apps didn't eliminate traditional private hire operators overnight.
Autonomous technology is likely to follow a similarly gradual path.
In October 2018, fully driverless private hire vehicles remain largely something passengers associate with technology demonstrations and science fiction.
Addison Lee and Oxbotica want to change that.
Their first task sounds surprisingly mundane: map the roads.
But behind that enormous mapping project lies a much bigger ambition.
A vehicle that can understand London.
Navigate its streets.
Collect passengers.
And take them to their destination without a conventional driver controlling the steering wheel.
Whether that can genuinely happen by 2021 remains uncertain.
But the direction of travel is becoming increasingly clear.
The next revolution in London's private hire industry may not be another booking app.
It may be the removal of the driver altogether.
Autonomous technology could eventually transform some airport journeys, particularly repetitive shuttle services between terminals, hotels and transport hubs.
For now, however, airport transfers remain one of the most demanding forms of private hire work.
London Airport Transfer provides pre-booked private hire journeys to and from Heathrow, Gatwick, Stansted, Luton and London City Airport, as well as airport-to-airport transfers and journeys between London's airports, hotels and residential addresses.
Airport passengers frequently require more than simply a vehicle that can navigate from one address to another.
Flights can be delayed.
Immigration and baggage reclaim are unpredictable.
Passengers may need assistance with luggage.
Families and groups may require larger vehicles.
International visitors may need help identifying their driver and finding their way through an unfamiliar airport.
That human element helps explain why conventional drivers are unlikely to disappear from airport transfers simply because autonomous technology becomes available.
In the future, passengers could have considerably more choice.
A driverless airport shuttle for a simple journey.
A shared autonomous vehicle for a lower-cost transfer.
Or a professionally driven private hire vehicle when personal service and assistance are required.
Addison Lee's 2021 target may prove ambitious.
But today's announcement provides a fascinating glimpse of how different London's airport transport industry could eventually become.
Written by London Airport Transfer for passengers travelling to and from London airports.