AN OLD joke among pilots asks: what do you need to fly a modern aeroplane?
The answer is a computer, a pilot and a dog.
The computer’s job is to fly the plane.
The pilot’s job is to feed the dog.
The dog’s job is to bite the pilot if he tries to touch anything.

It may be an exaggeration, but not by much: most long-haul flights are handled by autopilots from just after take-off
until right before landing. And if the airport has the necessary technology, even the landing can be handed to the computer.

Computerised control of transport is not new: the first autopilot, which allows a plane to maintain a steady course without
pilot intervention, was developed in 1912.

Two problems have been behind their stubborn no-show: one technical, the other psychological.
Computer-driven cars have proved technically more difficult than aeroplanes or trains because of the terrain on which
they travel. Aeroplanes spend much of the time in relatively empty skies, and the “stacks” they join while waiting to land
at airports are tightly controlled by human decision-makers on the ground.

Flying a plane is simple enough that many modern autopilots use Intel’s veteran 80386 processor—at 24 years old,
an antediluvian relic in computing terms, with less horsepower than the chip found in a modern mobile phone.

Trains, meanwhile, are conceptually simple: they can only move forwards or backwards, and most of the time drivers
need only watch for red signals and keep the train moving at the right speed.

Cars are more complicated because they must navigate a road system that is much more extensive and much less
standardised than a rail network. Roads are anarchic places compared with railways, which tend to be fenced off.
That helps to stop people or animals getting on to the tracks.

An automatic car would have to deal with all sorts of unexpected hazards, from accidents in other cars, to steering
clear of emergency-service vehicles, to stopping when a football rolls out into the road—with a child,
still hidden from view, in hot pursuit.

Yet engineers are still working on the problem because the advantages are so enticing. Once cars can reliably
sense hazards, they will react far faster than people. Communication between cars would allow traffic speeds to be
optimised, and avoid the wasteful overtaking and slowing down that people are so fond of
(and which helps, paradoxically, to cause traffic jams).

It could also revolutionise car design. With little need for human input, a car’s traditional layout could be abandoned in
favour of sofas, televisions, tables and even beds. That vision is slowly moving towards reality.
An €800m ($1.1 billion) European project made headlines in 1994 when a pair of driverless vehicles drove through
heavy Parisian traffic at speeds up to 80mph for 625 miles with only occasional human intervention.

A more recent competition to find the best computer-controlled car was organised by DARPA, America’s military
research agency. In 2007 a converted Chevy Tahoe navigated 40 miles of urban-style driving on an airbase in California,
obeying all the traffic laws, merging into traffic and averaging 14mph, to win a $2m prize. What may prove harder to
overcome is the psychological reluctance that people have to handing control to a computer, albeit one that is lightning
quick and tireless.

One of the reasons aeroplanes are able to be computerised so extensively is that passengers are unaware of when they
are being flown by the pilot and when they are controlled by a computer. Although it may seem odd to see no driver on a
train, passengers on trains are not used to being in control even when a driver is in charge, as is also the case with aeroplanes.

Yet even if people could be persuaded to give up their steering wheels for reasons of safety and comfort, many enjoy
driving their cars. It is hard to see how a computerised car can be reconciled with the idea of driving for pleasure.

But the slow rate at which the technological obstacles are being tackled could help overcome the psychological ones.
Many experts reckon fully autonomous, self-driving cars are two to three decades away, and that they will arrive through
lots of little changes rather than a single revolutionary invention.

The computers have already begun their takeover: computer-controlled cruise control, which uses laser or radar to maintain
distance from the car in front, has been available since the late 1990s. Several manufacturers, including Mercedes-Benz
and Volvo, offer cars that will brake automatically if they detect an imminent collision with a vehicle in front. The system is
advanced enough to handle cutting in from the side. Even if the driver does react in time, the computer will provide extra
power to the brakes to avoid a collision.

In 2007 Lexus, the luxury marque of Japanese car-giant Toyota, brought out a car featuring semi-automatic parallel parking:
the driver tells the car where he wants it parked and the machine does the rest. If the computer takeover happens slowly
and imperceptibly, it will be easier to learn to live with.