[Quietcars] New Scientist article on cars that drive themselves

[michael townsend]

FYI:  

Mike T in NJ


	WITH his jeans, white trainers and stripy top, Bob is every inch the
well-dressed 6-year-old. He's standing in the middle of a hotel car park
and, scarily, I'm driving straight at him. Instead of hitting the brakes, I
put my foot down on the accelerator. With just 10 metres to go, a row of red
lights flashes across my windscreen and there's an urgent, high-pitched
beeping sound. An instant later, I am jerked forward as the brakes slam on
automatically and the car screeches to a halt just short of Bob's stomach.
	
	This is what Bob is for. The child-sized dummy has just helped me
test the first in-car system that can sense an imminent collision with
pedestrians and brake automatically if the driver doesn't. It is being put
through final trials before being launched in May by Swedish car maker Volvo
in its new S60 model.
	
	The Volvo system is the latest in a line of developments made
possible by sophisticated sensors based on cameras, radar and lasers. These
sensors already provide drivers with adaptive cruise control, which alters a
car's speed to maintain a safe distance from the vehicle in front, as well
as technology such as semi-autonomous parking systems. Yet according to
Jonas Ekmark, a researcher at Volvo near Gothenburg, this is just the start.
	
	Ekmark says we are now entering an era in which vehicles will also
gather real-time information about the weather and highway hazards, using
this to improve fuel efficiency and make life less stressful for the driver
and safer for all road users. "Our long-term goal is the collision-free
traffic system," says Ekmark.
	
	Ultimately, that means bypassing the fallible humans behind the
wheel - by building cars that drive themselves. Alan Taub, vice-president
for R&D at General Motors, expects to see semi-autonomous vehicles on the
highway by 2015. They will need a driver to handle busy city streets or
negotiate complex junctions, but once on the highway they will be able to
steer, accelerate and avoid collisions unaided. A few years on, he predicts,
drivers will be able to take their hands off the wheel completely: "I see
the potential for launching fully autonomous vehicles by 2020."
	
	By about 2020 drivers will be able to take their hands off the wheel
completely
	Road traffic accidents kill about 37,000 people a year in the US and
39,000 in Europe, with driver error a contributing factor in over 90 per
cent of them. But a glimpse of a safer future has come from a trial,
completed in Sweden in 2008, of the Slippery Road Information System (SRIS).
The system used sensors and computers installed in 100 cars to gather
information on the use of brakes, fog lights, windscreen wipers and
electronic stability systems, as well as local weather conditions. Unlike
the Volvo system, in which each car uses only information from its own
sensors, the cars in the SRIS trial beamed the data they gathered to a
central database every 5 minutes.
	
	The study suggested that this pooled data could give drivers a far
more accurate picture of road conditions than local weather stations can.
Researchers still have to find the best way to merge this information and
broadcast it back to drivers. Nevertheless, the study concluded that
networks such as SRIS could improve safety and save lives.
	
	A more sophisticated system involving shared data is being deployed
in Japan this year. The country has become a world leader in the field
thanks to the government's decision to fund a network of infrared, microwave
and radio transmitters at the roadside.
	
	Around 2 million vehicles on Japanese roads can already pick up news
on congestion, roadworks, accidents, weather, speed limits and parking
availability from these transmitters, broadcasting as part of the Vehicle
Information and Communication System (VICS). Over the next few months,
cameras and sensors positioned around 20 major intersections in Tokyo and
Kanagawa prefecture will begin alerting drivers of cars with VICS receivers
to potential hazards such as vehicles attempting to merge into their lane,
or traffic crossing an intersection ahead. The new Driving Safety Support
System (DSSS), as the set-up is called, can also show alerts on satnav
displays warning of traffic lights, stop signs and even pedestrians and
cyclists on the road ahead. It will be in use at major intersections
nationwide by the middle of 2011.
	
	By that time, a similar system designed to operate on major Japanese
highways should have been running for a year. Called Smartway, it issues a
warning when the driver gets too close to the vehicle in front, when
vehicles are converging from the side, and when there is congestion ahead.
Some new vehicles from Nissan, Toyota and other car makers are already
equipped to use DSSS or Smartway. Older cars can access these systems too if
their receivers and satnav displays are upgraded. From here it is just a
small step - in technological terms, at least - to allowing cars to be
controlled automatically.
	
	Calling all cars
	
	In Europe and the US, vehicle manufacturers see direct
vehicle-to-vehicle communication as a simpler and cheaper solution than
building elaborate roadside infrastructure. Their plans envisage using Wi-Fi
links between vehicles to form ad hoc, reconfigurable networks that will
share information on road conditions, local weather and traffic accidents.
	
	The most ambitious of these projects, a collaboration between seven
European manufacturers and universities, aims to harness vehicle-to-vehicle
networks to make the driver redundant, at least for part of the journey.
Called SARTRE (Safe Road Trains for the Environment), it envisages up to
eight cars as little as a metre apart driving in convoy, controlled by a
lead vehicle operated by a professional driver.
	
	Ordinary drivers will book a place in convoys operating along major
roads. As they approach the convoy, they will hand over control of their car
to software on the lead vehicle. From then on, its steering, acceleration
and braking are controlled by an on-board computer that uses data sent
wirelessly from the lead vehicle, along with information from cameras and
radar and laser detectors on the front and rear of the car itself. Drivers
will be able work, read, watch films or even sleep while their cars are
driven for them. "It will be like sitting on a bus or a train," says Ekmark.
But when the convoy nears an exit at which drivers wish to leave, they can
resume control and continue their journey.
	
	As well as being protected against collisions, cars in a convoy use
less fuel than when they are travelling separately, and they take up less
road space. At highway cruising speeds, aerodynamic drag can be reduced by
as much as 60 per cent when vehicles are separated by less than one car
length. Overall, convoys are predicted to cut fuel use and carbon emissions
by up to 40 per cent.
	
	Unlike a previous generation of car trains developed at the
University of California, Berkeley, during the 1990s, SARTRE convoys will
run on public roads alongside ordinary traffic. The Berkeley project failed
to get off the ground because it required specially built roads, making the
concept prohibitively expensive. If this year's trials of SARTRE planned for
test tracks in Sweden and the UK are successful, a full demonstration -
consisting of a lead truck followed by another truck and three cars - is
planned for public roads in Spain towards the end of 2011. Before that can
happen, however, the SARTRE consortium must work out how a convoy will
interact with other road users. For instance, will it have to break up when
overtaking, and then reform once all its members have passed the slower
vehicle?
	
	The long journey towards cars that will drive themselves began in
1971 with anti-lock brakes. "That was the first time we introduced the
overriding of driver input," says Taub. Another step along the road came
with electronic stability control, which governs brakes, steering and
throttle to prevent cars going off the road in an uncontrollable skid.
Top-of-the range cars are increasingly being fitted not only with adaptive
cruise control but also with lane assistance, which gently applies the
brakes to keep cars from straying out of lane.
	
	Taub expects these systems to start appearing on cheaper models over
the next few years. "We still have the driver in the loop with eyes on the
road, hands on the wheel, feet on the pedals," he points out. "But
increasingly the vehicle will be steering and accelerating on its own."
	
	What fully autonomous vehicles will be like is hinted at by an
experimental car called Boss. Built by a team of engineering students at
Carnegie Mellon University in Pittsburgh, Pennsylvania, and backed by
General Motors, this robotic car scooped a $2 million prize by outperforming
10 other autonomous vehicles in a simulated urban environment created for
the DARPA Urban Challenge in 2007. To win, Boss had to execute complex
manoeuvres such as merging into flowing traffic, overtaking, parking and
negotiating intersections, while interacting with other autonomous vehicles
and 30 human-driven ones.
	
	Boss's computer builds a model of the immediate environment by
processing data from radar, laser sensors, cameras and GPS. It then uses
this model, along with information such as local traffic rules, to plan the
best route and provide the situational awareness the vehicle needs for
manoeuvres such as changing lanes safely, or to determine whether it has
priority at an intersection.
	
	Boss uses sensors and other components that are already fitted in
production vehicles, but the computing power it uses to handle all the data
is a different matter. It currently requires the equivalent of 10 desktop
computers, and miniaturising the electronics so that it can be hidden away
in a normal-sized car remains a challenge. Another task will be to develop
the interfaces between car and driver and find simple ways to switch control
from manual to automatic and back again.
	
	Taub predicts that by about 2020 vehicles like Boss will start to
appear on public roads; drivers will be able to disengage totally and hand
control over to the car. "You'll see a progression of subsystems, with costs
coming down and increased robustness," he says.
	
	At Stanford University in California, the Volkswagen Automotive
Innovation Lab has shown what might be possible. VAIL engineers have fitted
a VW Passat with cameras, cruise control radar and laser sensors, allowing
it to navigate a parking lot, spot an empty space and park perfectly, with
or without a driver.
	
	Manoeuvring at low speed is one thing, but are we ready to hand over
control on the open road? How would you feel about being at the mercy of a
machine barrelling along the highway at 100 kilometres per hour or more,
with your family in the back and you merely a passenger at the wheel?
Confidence in the reliability of electronic drive-by-wire controls took a
knock in January when Toyota had to recall millions of its vehicles. A few
accidents involving autonomous vehicles could set the whole idea back years.
	
	Automated manoeuvring at low speed is one thing, but are we ready to
hand over control on the open road?
	Though advances in communications and connectivity have transformed
our world, it is still not easy to envisage a highway network populated by
cars that drive themselves more safely than any human can. Yet if Ekmark and
Taub are right, the next generation of vehicles will be able to do just
that. The real question may be whether we will have the nerve to take our
hands off the wheel and let the machines take over.
	
	Early adopters
	Who wants to pay to be first with a technology that only works when
lots of other people already have it? That is likely to be the big problem
facing car-to-car networks once the technical questions have been sorted
out.
	
	One way to minimise this problem is to make the equipment cheap to
retrofit into existing vehicles. General Motors has demonstrated a system
called V2V, which costs less than $200 to install. It uses GPS and Wi-Fi to
warn drivers of hazards such as vehicles in blind spots.
	
	Others see the cellphone network as the key. Cellphone operator
Orange is one of six UK organisations in a partnership called Sentience,
which is developing a low-cost system based on GPS-enabled smartphones. The
system acquires and combines information from topographical maps and traffic
data in order to control a vehicle's brakes and accelerator. In tests, the
Sentience system reduced fuel consumption by up to 24 per cent over that of
a car driven normally.
	
	Another approach was highlighted at the Cooperative Mobility
conference in Amsterdam, the Netherlands, in March, when the European
Cooperative Vehicle-Infrastructure Systems (CVIS) consortium showed off its
universal communications system. This allows vehicles to swap information
with each other and with networks using 3G, GSM, infrared or wireless
protocols, and to switch seamlessly between these modes. CVIS is providing
developers with kits to help them create services to run on its
open-architecture platform.
	
	Later this year CVIS plans to unveil an in-car touchscreen
applications unit. CVIS coordinator Paul Kompfner envisages a
smartphone-like interface that will offer drivers a range of apps depending
on their location. One app under development communicates with traffic-light
control systems and tells drivers what speed they should travel at to pass
without hitting red.
	
	Another app, to be tested later this year in Poland and the
Netherlands, allows trucks to take priority by controlling traffic lights as
they get near. "If you give priority to trucks it is not just the trucks
that gain - overall traffic efficiency and flow are improved," says project
manager Zeljko Jeftic.
	
	Nic Fleming is a science and technology writer based in London