Automotive

Volvo's autonomous driving trial will be UK's "most ambitious"

Volvo's autonomous driving trial will be UK's "most ambitious"
Drive Me London is due to start early next year, with members of the public testing "a limited number" of semi-autonomous cars
Drive Me London is due to start early next year, with members of the public testing "a limited number" of semi-autonomous cars
View 5 Images
Drive Me London is due to start early next year, with members of the public testing "a limited number" of semi-autonomous cars
1/5
Drive Me London is due to start early next year, with members of the public testing "a limited number" of semi-autonomous cars
The Drive Me London trial will be expanded in 2018 to include up to 100 fully autonomous cars
2/5
The Drive Me London trial will be expanded in 2018 to include up to 100 fully autonomous cars
Volvo says it believes the introduction autonomous cars will "revolutionize" Britain's roads
3/5
Volvo says it believes the introduction autonomous cars will "revolutionize" Britain's roads
Some of Volvo's cars already feature semi-autonomous driving technologies
4/5
Some of Volvo's cars already feature semi-autonomous driving technologies
The Drive Me London trial is aimed at producing data that can be used to develop fully autonomous cars
5/5
The Drive Me London trial is aimed at producing data that can be used to develop fully autonomous cars
View gallery - 5 images

Having announced what it calls China's "most advanced autonomous driving experiment" earlier this month, Volvo has now revealed plans for the UK's most ambitious autonomous driving trial as well. Drive Me London will see real families driving autonomous cars on public roads.

Volvo believes that the introduction of autonomous cars will "revolutionize" Britain's roads. In particular, it expects roads to be made safer, less congested, less polluted and less time consuming to get around.

The reasoning is that the data produced by sensors used on autonomous cars and their accompanying computer control systems are said to allow them to better monitor other vehicles around them and adjust their speed accordingly. This, in turn, means that traffic flows more smoothly, reducing traffic jams, emissions and congestion as a result.

The carmaker also says research suggests autonomous cars will have the potential to reduce accidents by up to 30 percent, with up to 90 percent currently caused by driver error or distraction. This is a view shared by the chief executive of trial partner Thatcham Research, Peter Shaw.

"Without doubt, crash frequency will also dramatically reduce," says Shaw. "We've already seen this with the adoption of Autonomous Emergency Braking (AEB) on many new cars. Research in the US by NHTSA predicts that by 2035, as a result of autonomous and connected cars, crashes will be reduced by 80 percent. Additionally, if a crash unfortunately can't be avoided, then the impact speed will also drop as a result of the system's performance, reducing the severity of the crash."

Volvo says it believes the introduction autonomous cars will "revolutionize" Britain's roads
Volvo says it believes the introduction autonomous cars will "revolutionize" Britain's roads

The Drive Me London trial is aimed at producing data that can be used to develop fully autonomous cars suitable for use in the real-world. Volvo says that such data can help to do so in a way that "more unrealistic" data produced on test tracks cannot.

Thatcham Research's role in the trial will be to provide technical analysis of the data produced and, where required, to provide professional test drivers.

Drive Me London is due to start early next year, with members of the public testing "a limited number" of semi-autonomous cars. The plan is then for the trial to be expanded to include up to 100 fully autonomous cars in 2018.

Source: Volvo

View gallery - 5 images
4 comments
4 comments
KirkAugustin
Just stop it! No one is going to want to ride in an autonomous car. They will never be safe. The current ones are fake, relying on GPS mapping, which is not at all reliable enough. The Google car does not even recognize turn signals, so ran into a bus signaling its move back into the moving lane. Lidar will never be able to distinguish a child from blowing paper. Terrorists could exploit autonomus cars. Every accident will sue the maker. What could an autonomous car do when snow covers the markers? The whole idea is just crazy hubris like the 1950s "flying cars".
Stephen N Russell
Test in So CA area, then come to the US after UK.
Jack Decker
Here's what Kirk gets wrong.
1) First, there is no "GPS mapping". Do autonomous cars use GPS maps, yes. Do they create GPS maps, no. Autonomous vehicles use a combination of radar and laser scanning (and some are incorporating infrared sensors) which is cross-checked with internally stored maps. Using GPS to locate the car on the map is simply a supplementary confirmation. In other words, an autonomous car doesn't need GPS to drive roads, is designed NOT to depend it, and cannot use it when driving through things like tunnels.
2) While the Google car can recognize vehicle turn signals, in the accident with a city bus, the city bus didn't use any turn signal. Oh, the Google car can also identify hand signals by cyclists and traffic cops.
3) In the accident, the city bus didn't yield way as road rules and laws require vehicles to do. The Google car was driving around an obstruction (sand bags around a street drain), waited for other cars to pass, and then tried to pass the obstruction thinking the city bus would do what it was supposed to do and yield to a merging car. The bus didn't. It is still debated who was at fault.
4) Kirk's assertion that the Google car ran into the bus is hyperbole. If anything, it walked into it. The crash was at speeds 2 mph for the Google car and 15 mph for the bus. No injuries. Literally only a dented fender on the Google car and no damage to the bus. In the internal video of the bus, it is clear that passengers weren't even nudged in their seats.
5) I assume Kirk meant "laser" and not "lader". And, yes, they can distinguish between a child and blowing paper. Google (or Bing if you don't trust Google) the phrase "how Google car sees the world" and you'll see the very detailed laser images the cars generate and use. In fact, you'll likely turn up a TED Talk video by Chris Urmson that shows and explains just how the Google car sees the world. Well worth the watch.
6) Will terrorist exploit autonomous cars? Yes. They also presently exploit non-autonomous cars. If anything, terrorists will use autonomous vehicles less since part of the reason for the suicide car bombing is the terrorist actually committing suicide to shock the public that they'd actually die for their cause. That's true with any suicide bombing. You see there is this brand-new technology called "timers" (think of them as clocks that ring when they hit a time mark) that these bombers could now use that makes them having to personally push the detonation button unnecessary. Yes, yes, I know. Timers are high-tech science fiction stuff but, honestly, they don't cost more than a few thousand dollars each. In fact, I think you can get them for a lot less at Wal-Mart.
7) Yes, auto-makers will be sued for auto accidents caused by the autonomous vehicles they sell. After all, there will be no human driver in control of the car. The car will be in control of itself. He who controls the car is responsible. But here's the thing. The most unreliable part of the car is the human driver. Over 95% of all car accidents are due to human error. Remove the human and we are expected to see at least an 80% reduction in car accidents. And here's another thing. Unlike human drivers, the autonomous cars will be constantly improving their driving. Every accident that does occur, the auto makers will download the internal recorded driving record of the accident and learn from and program the cars how to better handle those situations. And even when there is an accident, the autonomous car can brake and steer in the best way possible to reduce the severity of the accident. That will reduce injuries and deaths. Unlike the human driver, the autonomous car won't scream, panic, and do something stupid or not do anything but brace for impact. Instead, the autonomous car will follow its programming perfectly right to the point of impact and beyond. Oh, and the autonomous cars already think at least 1,000 times faster than humans. To an autonomous car, it is thinking so fast that the world it moves through is almost standing still. Or to put it another way, take any car accident. Any. Even ones done on a NASCAR race track at full speed and then slow everything down by 1,000. The fastest NASCAR race car speed record is 200.11 mph. Divide that by 1,000 and if you were an autonomous car, it would be like you the human were driving that superfast NASCAR at .20011 miles per hour. How fast is that? The speed of a slug.
8) As for snow covering the road, no problem. I'll do you one better. No paved roads. Autonomous cars can and have driven off-road. They've driven in all conditions. Snow, hail, rain, high winds, blowing sand, extreme heat with surface mirages in effect, without headlights on, etc. Oh, and headlights on an autonomous car will be only turned on for the human passengers. Autonomous cars don't need them. So when it comes to a snow-covered road, that's where the internal maps of the autonomous car come into play and, yes, GPS assisting. Unlike the human driver, in white-out conditions, the autonomous car "sees" just fine. In fact, it sees better than fine. Unlike humans, it sees 360 degrees around itself continuously. No blind spot. No just forward vision with glances at rear view mirrors. Autonomous cars don't use rear view mirrors since they are always looking in all directions at all times.
Mel Tisdale
It Is worth remembering that autonomous aircraft technology has been around for many years, yet pilotless passenger aircraft have yet to see the light of day. Why? I think we can safely put that down to psychology and a built-in mistrust of machinery. The same is going to apply to autonomous cars. Yes, I am sure that in everyday circumstances an autonomous car will be much safer than one driven by a human being, five minutes on YouTube will confirm that. BUT what is to be done about our friend Taleb and his black swans? What happens when a ladder falls off the back of a builder's truck? Or a massive sinkhole opens up in the road, or a bridge gets washed away? I suppose we might get lucky and have them happen together so we can use the ladder to rescue the poor people whose clever cars have driven into the sinkhole. We had better hope that those who have plunged into the river can swim. The ladder might not be too much of a problem as long as the technology can recognise it. But the sinkhole is different. Unless the vehicle can come to a complete stop before it reaches it, serious injury will result. O.K. we could have them drive more slowly, but then we will have increased congestion.
My biggest problem with autonomous road vehicles is that they are leap-frogging a vital stage in their development. They need to have the autonomous technology in place as enhanced driver assistance first and have an alert human being at the helm. The vehicle should warn the driver if they are about to have accident if they continue to deviate from the previously safe path that they were on, and take over as a last resort (recording the error in the process. The vehicle must not be allowed to break any speed restrictions, even the most temporary or most recent. Also, it must recognise stop signs etc. and traffic-light signals, ensuring that they are obeyed in the process.
In engineering we have a saying: 'If it can happen, it will happen.' Sensors can fail, so they will fail. What then? Leave a trail of injured cyclists and bent bodywork behind? But if the car is to be rendered inoperable for the sake of the cyclists and other cars, what about the thief who has painted over the sensor, or glued some aluminium foil over it, so that they can rob the stranded owner? Or what about the fact that on a dry road autonomous cars will all perform to a similar standard? Robber number one steps out in front, car stops just in time, robber number two steps behind it, thus the car either runs someone over, or it cannot move. While it is pondering the dilemma in super quick time, robber number three smashes the windows in and robs the occupant(s).
On top of all this we are going to have to cope with an aging fleet that is going to take 20 to 30 years to reach the breaker's yard. Wouldn't it make sense to think in terms of enhanced driver assistance for all new vehicles so that they can be ready for full autonomous use when (if) deemed safe?