Peugeot Citroën's self-driving car makes Paris to Bordeaux run

 

An autonomous Peugeot Citroën has driven itself all the way from Paris to Bordeaux in a demonstration of the increasing viability of self-driving vehicles (Credit: PSA Peugeot Citroën) A PSA Peugeot Citroën driverless vehicle has journeyed 580 km (360 miles) on the motorway from Paris to Bordeaux entirely in autonomous mode. Automatically maintaining its speed to the road conditions and traffic, as well as independently changing lanes to pass slower traffic, the Peugeot Citroën completed the journey without incident and demonstrated the increasing viability of driverless motor vehicles. Motoring to Bordeaux to participate in the Intelligent Transport Systems (ITS) World Congress (5 to 9 October), the car was in attendance to show off PSA Peugeot Citroën’s new autonomous technology along with its newly-developed car-to-car and car-to-infrastructure communication systems. The on-board sensors and car and infrastructure communications demonstrated are claimed to provide a raft of safety features, including pedestrian awareness and collision avoidance from information transmitted by cars ahead and via dynamic road signs designed to deliver tailored messages to drivers. The autonomous vehicle itself contained a vast array of on-board systems, including radar, automated braking, steering, and acceleration, GPS, and digital cameras, to name but a few. All working in concert with the master electronic control unit – and monitored remotely by a Peugeot Citroën control center – the autonomous vehicle is designed to be safer than driver-only vehicles, where hazards such as fatigue and human error are a major factor in collisions. "The journey made by our prototype today proves that autonomous vehicles are no longer a matter of science fiction," said Carlos Tavares, Chairman of the managing board of PSA Peugeot Citroën. "This ushers in a new era of mobility, which I find truly exciting." Much like the Mercedes-Benz autonomous truck recently tested on a stretch of Autobahn in Germany, the PSA Peugeot Citroën vehicle was permitted to drive on the highway with stringent protocols in place. With a driver ready to take control at any time, the vehicle's degree of autonomic control was set at what is known as "Level 3"; where the the vehicle is able to accelerate, brake, steer, and control the vehicle. Peugeot Citroën also included what it calls the Scoop@F system, which allows not only for the autonomous controls and mechanisms mentioned earlier, but a range of more intuitive vehicle controls. Such things include adaptive cruise controls that manages vehicle speeds as standard systems do, but also allows for upcoming traffic light changes, hills, obstacles, and traffic flow via intrinsic communications systems collecting data from other vehicles and traffic infrastructure. Unlike many autonomous concepts, however, PSA Peugeot Citroën says its vehicle will be fully-capable of "Level 5" autonomous control (subject to legislative approval) in the not-too-distant-future. Slated for release to the public by 2018, Peugeot Citroën customers will be able to purchase vehicles fitted with all the features demonstrated in this most recent test run (aka "Level 3") which, the company believes, will greatly enhance customer safety by relieving the driver of fatigue from the tedium of long-distance driving or inattention.   Source: PSA Peugeot Citroën http://www.gizmag.com/peugeot-citroen-autonomous-car-paris-bordeaux/39738

Google’s autonomous cars will drive themselves around Mountain View this summer

 

Newsfeeds have been abuzz lately with discussions about the safety of Google’s self-driving cars. On the heels of the news that none of the autonomous vehicles were at fault in the motor vehicle ‘incidents’ they’ve been involved in, Google’s Self-Driving Car Project plans to free the cars from the confines of their test track and let them loose this summer on the streets of Mountain View, California. Well, sort of. In a blog post, Chris Urmson, director of the Google Self-Driving Car Project, explains that the cars will head out onto the open road with a safety driver on board. One of the biggest goals of the self-driving car movement is to free up drivers to do other tasks instead of, well, driving. Google’s autonomous car, like many other emerging prototypes, still has a steering wheel and brakes so a human driver can take over if necessary, but the car’s ability to drive itself is ever-improving. In a press conference, Google co-founder Sergey Brin said the search engine giant’s self-driving vehicles are capable of recognizing police cars, joggers, hand signs from traffic cops, and the inevitable fact that “occasionally people make terrible decisions—turning right from the far left lane, for instance, or running a red light—and is on alert for them.” Related: Google unveils the first fully functional prototype of its self-driving car By putting the autonomous cars on the road, Google’s engineering team aims to learn even more about how the vehicles operate and play with other traditionally operated cars (i.e. human-driven cars), as well as the host of variables that drivers encounter on a daily basis. The team is also really keen on getting people excited about the technology, by allowing them to see and interact with the cars. From the announcement, it doesn’t appear that Google plans to roll out all 23 of the self-driving prototype cars—just “a few.” The models that will be tooling around Mountain View have “the same software that our existing fleet of self-driving Lexus RX450h SUVs uses,” according to Urmson’s blog post. “That fleet has logged nearly a million autonomous miles on the roads since we started the project, and recently has been self-driving about 10,000 miles a week. So the new prototypes already have lots of experience to draw on—in fact, it’s the equivalent of about 75 years of typical US adult driving experience.” I guess that explains why they haven’t been to blame for any of the accidents. + Official Google Blog Images via Google   http://inhabitat.com/googles-autonomous-cars-will-drive-themselves-around-mountain-view-this-summer/

Google self-driving car involved in first injury accident

 

Fri, 07/17/2015 - 1:00pm Justin Pritchard, Associated Press In this May 13, 2015, file photo, Google's self-driving Lexus car drives along street during a demonstration at Google campus in Mountain View, Calif. Google says that one of its self-driving cars has been involved in an injury accident for the first time. The tech giant disclosed Thursday, July 16, 2015, that one of its SUVs was rear-ended in its home city of Mountain View, and the three people on board complained of minor whiplash. All were released from the hospital soon after the July 1 collision. (AP Photo/Tony Avelar) Google Inc. revealed Thursday that one of its self-driving car prototypes was involved in an injury accident for the first time. In the collision, a Lexus SUV that the tech giant outfitted with sensors and cameras was rear-ended in Google's home city of Mountain View, where more than 20 prototypes have been self-maneuvering through traffic. The three Google employees on board complained of minor whiplash, were checked out at a hospital and cleared to go back to work following the July 1 collision, Google said. The driver of the other car also complained of neck and back pain. In California, a person must be behind the wheel of a self-driving car being tested on public roads to take control in an emergency. Google typically sends another employee in the front passenger seat to record details of the ride on a laptop. In this case, there was also a back seat passenger. Google has invested heavily as a pioneer of self-driving cars, technology it believes will be safer and more efficient than human drivers. This was the 14th accident in six years and about 1.9 million miles of testing, according to the company. Google has said that its cars have not caused any of the collisions - though in 2011 an employee who took a car to run an errand rear-ended another vehicle while the Google car was out of self-driving mode. In 11 of the 14, Google said its car was rear-ended. In a blog posted Thursday, the head of Google's self-driving car program, Chris Urmson, wrote that his SUVs "are being hit surprisingly often" by distracted drivers, perhaps people looking at their phones. "The clear theme is human error and inattention," Urmson wrote. "We'll take all this as a signal that we're starting to compare favorably with human drivers." In a telephone interview, Urmson said his team was exploring whether its cars could do something to alert distracted drivers before a collision. Honking would be one possibility, but Urmson said he worried that could start to annoy residents of Mountain View. According to an accident report that Google filed with the California Department of Motor Vehicles about the July 1 crash: Google's SUV was going about 15 mph in self-driving mode behind two other cars as the group approached an intersection with a green light. The first car slowed to a stop so as not to block the intersection - traffic on the far side was not moving. The Google car and the other car in front of it also stopped. Within about a second, a fourth vehicle rear-ended the Google car at about 17 mph. On-board sensors showed the other car did not brake. The driver of that car reported "minor neck and back pain." The SUV's rear bumper was slightly damaged, while the vehicle that struck it lost its front bumper. Mountain View police responded, but did not file an accident report. Source: Associated Press

Google’s Homemade Self-Driving Cars to Hit Roads This Summer

 

Google Here’s something that irks Chris Urmson: Sometimes people will get in self-driving cars, the spectacularly complex piece of technology he runs at Google and to which he has devoted most of his scientific career, and leave with a shrug. Once, Urmson was riding in one of Google’s Lexus SUVs down a freeway. Several minutes in, his fellow passenger turned to him, nonplussed. “That’s it?” Urmson, recalling the story on Google’s Mountain View campus earlier this week, threw up his hands: “Do you have any idea how hard this is?!” Soon, there may be many more blasé reactions to one of Google’s most audacious moonshots. On Friday, the Internet giant announced that the first autonomous vehicle it has manufactured — a squat two-seater, unveiled a year ago, with no steering wheel or brakes — will begin rolling out on public roads in northern California this summer. Urmson and his team have assembled 25 of the cars, which, for now, are just called “prototypes.”   When they hit the roads, they will not exceed 25 miles per hour. And, due to current state regulations, they must be equipped with brakes, an accelerator pedal and a steering wheel. But ultimately, Google wants to strip those out. The company’s stated goal is shepherding fleets of vehicles that can drive with no need for human intervention, a bid to curtail the time wasted in traffic and aide those unable to drive. “At that point, the steering wheel and brake pedal just don’t add value,” Urmson said during the demonstration at the new Google X headquarters in Mountain View. “Over the last few years, we’ve been focused almost purely on tightening the technology. The big next step is bringing it into the community and seeing how it mixes with people.” At its event to show off the car, Google mixed with people. Along with press, Google invited local community members and disability advocates onto its spacious, secure rooftop. Sergey Brin, the co-founder behind Google’s futuristic ventures, made an appearance. He reiterated Urmson’s point, arguing that for one target market of the technology — several blind community members were given lifts in the car — steering wheels aren’t the issue. “That doesn’t address the mission of access,” Brin said. Google’s announcement comes on the tail of sharp criticism. On Monday, the Associated Press reported that Google’s Lexus vehicles were involved in three accidents since September, when California required autonomous vehicle testers to declare a permit. Later that day, Google released (incidentally, by Google’s telling) the first glimpse at numbers on its self-driving car experiment: 11 accidents during the 1.7 million miles on the road since 2009. (That puts its incident rate at more than twice the national average of 0.3 damaging incidents per 100,000 miles.) Urmson detailed the accidents: Seven came from other cars rear-ending theirs, two were freeway side-swipes and one was a silly error from a Google test driver who was using the manual controls at the time. Google insists that the higher rate comes from thorough reporting, something most human drivers ignore. When its homemade cars hit the pavement, Google will also launch a website for community feedback on the trials, and will begin posting regulator progress reports, including miles driven, noteworthy trends and incidents. “It sounds cliche, but safety is issue one, two and three,” Brin told the audience. Google’s built-from-scratch car looks similar to its debut a year ago, when Re/code took it for an inaugural spin. It uses the same complex software and hardware — a jury-rigged, advanced network of swirling lasers, cameras and radar — as the existing Lexus fleet. Over the last year, the cars have grown considerably smarter and more adept, said Dmitri Dolgov, who leads software for the self-driving cars. They can decipher a trash can from a pedestrian, and even pick up what a pedestrian’s hand motions mean. Gradually, they’re also learning to handle unusual traffic situations. In Mountain View, Urmson showed earlier footage of Google’s Lexus at an intersection when a renegade cyclist crossed in front, running a red light. As the light turned, a truck to the Lexus’ left veered ahead, barely missing the cyclist. Google’s SUV saw it and stood still. (One car encountered something rarer still: A wheelchair-bound suburbanite chasing a duck; the car opted to stall.) If anything, the car errs on the side of caution. At the Google X headquarters, Google offered rides to the select few community members and reporters. During my ride, the car easily handled the planned obstacle course. A gentle slowdown when a Googler suddenly walked in front. A smooth turn when another veered ahead on a bicycle. But when my car turned to face the unexpected gaggle of press surrounding Brin, it jolted to a halt. Then lurched ahead like a nervous 16-year-old. The car is not accustomed to large gatherings of people in open spaces, Dolgov explained. Mark Bergen Sergey Brin, at Google X headquarters Yet, it learned from the encounter. With each ride, the cars deposit the observed data and share it across the entire fleet. Also, Google has learned more political savviness. The California DMV has awarded testing permits to Google and six other manufacturers, including Mercedes-Benz, Nissan and Tesla. But industry observers said the company has advanced more aggressively in lobbying. In Nevada, which granted Google the state’s maiden self-driving license in 2012, Google was the driving force in the policy process, at the expense of rivals. “The DMV and especially the state legislature, only listening to Google, wrote a law that was fine for Google but was really problematic for car manufacturers,” said Ryan Calo, a law professor and robotics specialist at the University of Washington campus. On the Google X campus, Brin, outfitted in shorts and Crocs (but no Glass), offered some boilerplate executive-speak. (“We are still refining our business plan.” “The regulatory issues are non-trivial.”) But he also hinted at the ambition of the program. “We’ve had pretty good conversations with a number of states,” he said. “And, for that matter, a number of countries.” Someone asked about his declaration, in 2012, that his self-driving cars would be ready for public use in five years. “That’s still right on track,” he said, before turning to his auto director. Urmson sheepishly corrected him — it’s closer to five years from now. Brin, whose mathematics prowess built Google’s search engine, replied: “Well, I haven’t done the math.”

Teens probably won’t like self-driving cars, but their parents will

 

If consumers have their way, self-driving cars will enable parents to keep tighter reins on teen motorists. A survey conducted by the College of Engineering at Carnegie Mellon University reveals that people are soundly in favor of putting parental controls in high-tech cars of the future. One thousand people, aged 18 to 70, were polled to learn which freedom-foiling attribute they deemed most important. Top Parental Controls Control to set speed limit, curfew time and number of passengers (84 percent) Control feature to limit the geographic range the car will travel (61 percent) Parent text display to communicate with driver (60 percent) Roughly 84 percent all respondents wanted to control: a car's speed, the number of friends who can pile into the car and the driver's curfew time. Women (87 percent) were strongly in support of this capability, as were 91percent of people aged 66 to 70. Even 81 percent of the youngest polled, ages 18-24, favored these novel features. Implementing these types of control technologies could save lives, prevent injuries and reduce costs associated with accidents. In 2013, 2,524 teenagers perished in motor vehicles crashes, making vehicle accidents the leading cause of death for teenagers. Compared to older drivers and miles driven, teen drivers are three times more likely to be in a fatal wreck. Young, inexperienced drivers tend to speed and drive too fast for road conditions. Further, teens are more likely to crash when they have teen passengers in the car. When it comes to curtailing the distance teen drivers can travel, men (62 percent) and women (61percent) closely align on this point. This notion, however, did not resonate well with 18- to 24-year-olds. Only 54 percent of them opted for this feature, whereas nearly 65 percent of drivers aged 36 to 45 would constrain a car's geographic range. The one area where 18- to 24-year-olds outscored all other age groups was in their receptiveness to having a parental text display in the car. Surprisingly, 69 percent of the youngest respondents thought this was useful while only 53 percent of people aged 56 to 65 would consider this option. Women (63 percent) tended to be more receptive than men (57 percent) to this communication feature. About the survey: Carnegie Mellon, the birthplace of autonomous vehicle technology, has a 30-year history of advancing self-driving car technology for commercialization. The college polled 1,000 people to gain insight into what consumers are looking for in self-driving cars. In the survey, a self-driving car was defined as having sensors and computing technology that allows the car to safely travel without a driver controlling the steering wheel, gas and brake pedal. The vehicle would automatically move at safe speeds, keep a safe distance from surrounding cars, change traffic lanes, obey traffic signals and follow GPS directions to destinations. Story Source: The above story is based on materials provided by Carnegie Mellon. The original article was written by Sherry Stokes. Note: Materials may be edited for content and length.

The pros and cons of a driverless future

 

Despite obvious advantages, the era of autonomous vehicles may not be all plain sailing (Photo: Shutterstock) It may seem as if the only arguments against self-driving cars come from two kinds of people – those fearful of any scenario where they might have to forgo control behind the wheel and those who distrust all technology out of hand. Lolling about while a computer gets you through traffic has its attractions for many of us and there has been little discussion about the potential downsides of a driverless future, but a new study has pointed out some potential flaws in this looming auto utopia. The upsides seem pretty self-evident. The first advantage of driverless cars is one of multitasking. While being taken from here to there in a driverless car, you can do anything you want. Eat, sleep, work, chat with relatives – commute time is no longer down time. Then there's safety. Although this has yet to be born out, the theory goes that people, on the whole, are not the best drivers out there. We might all think we're Mario Andretti behind the wheel, but sadly and all too obviously, we are not. By leaving the driving to a whole slew of computers, sensors, servos and software, getting from home to office should be rendered accident free. The other potential plus is efficiently. While a driverless car is using its digital prowess to whisk you to the store, it can, and will, be talking to all the other driverless cars out there. And not only chatting with them, but talking back and forth with smart roadways. This will allow all sorts of efficiencies to be realized. One driverless car will know that the other driverless cars in front of them will not suddenly slow down, so the gaps between can be shortened. And the first driverless car in this line, when it realizes that it will have to slow down, or stop, will relay this information to all the other driverless cars in the line, and they can react as a unit. Intelligent roadways will allow the timing of traffic lights to be optimized to the known level of traffic density and speed. Swoosh, off you go! From home to office to store and back home again with nary a red light to be seen along the way! Scott Le Vine, Alireza Zolfagharib, and John Polak have a different take. The researchers from the Department of Geography at SUNY in New Paltz, New York and at the Centre for Transport Studies, Department of Civil and Environmental Engineering, Imperial College London, have recently published a study that, while not slamming on the brakes for self-driving cars, does point out that it might not all be a smooth road ahead. "Autonomous cars are expected to increase road network capacity and reduce the disutility of travel time," but " ... under certain circumstances these benefits may be in conflict." The study examined how two of the three stated benefits – more comfort and less traffic – can live together by running a computer simulation of how driverless cars would work in practice, and comparing that to trains. Commuter trains are used in many the same ways as driverless cars; you get in, sit down, and read a book or do some work on your commute. Trains work very well for this because trains are smooth. Once they are up and running, they are relatively much more comfortable environment over the stretch of a commute than cars, even driverless ones, could potentially be. It is that potential problem that the study points to. If driverless cars do a lot of stop/start work like normal cars, even at a reduced velocity, they will become a less workable environment. But, if they speed up and slow down more smoothly for the passenger's sake, essentially mimicking the acceleration and deceleration of trains, then self driving cars would forfeit much of their capacity to alleviate congestion in the process. "Acceleration has big impacts on congestion at intersections because it describes how quickly a vehicle begins to move," says lead researcher Scott Le Vine. "Think about being stuck behind an 18-wheeler when the light turns green. It accelerates very slowly, which means that you're delayed much more than if you were behind a car that accelerated quickly." The team replicated traffic at a run of the mill four-way urban intersection where 25 percent of the vehicles were driverless. In some simulations, the driverless cars traveled the way that light rail trains do, gently on the gas then gently on the brakes. In other words more comfortable than a normal car, but still herky-jerky at times. In other scenarios, the driverless rides started and stopped like a high-speed rail train: smoothness in all things. Le Vine and colleagues tinkered with a number of parameters such as longer yellow lights or following distances. They modeled 16 scenarios against a control group with all human-driven cars, running each simulation for an hour and repeating it 100 times. They considered the normal influence each situation had on traffic with regards to delays and road capacity. The bottom line was this: In every single test, self driving cars where calculated to create a comfortable, rail-like ride made congestion worse than it would have been in a baseline scenario with people behind every wheel. When driverless cars accelerated and decelerated in the style of light rail, the congestion deteriorated from 4 percent to 50 percent and the number of cars traveling through the intersection also fell between 4 percent and 21 percent. Going for high speed rail style of smoothness, those numbers got even worse: Delays increased from 36 percent to nearly 2,000 percent and intersection capacity fell between 18 percent and 53 percent. "Our findings suggest a tension in the short run between these two anticipated benefits (more productive use of travel time and increased network capacity), at least in certain circumstances," say the researchers. "It was found that the trade-off between capacity and passenger-comfort is greater if autonomous car occupants program their vehicles to keep within the constraints of HSR (in comparison to LRT)." It is also worth pointing out that these simulations just involved normal cars and driverless cars. There were no trucks, buses or, that even more chaotic element, pedestrians. As a single simulation-based study, the research may not add up to a crippling blow for driverless car advocates, but it shows that there's work to be done in order find a liveable recipe for the coming mix of autonomous and conventional vehicles on our roads. Source: Autonomous cars: The tension between occupant experience and intersection capacity (Science Direct) via The Atlantic.

Google unveils first complete self-driving car prototype

 

Google has unveiled the first complete prototype of its self-driving car Google has been developing autonomous car technology since 2010. Now, it has unveiled the first complete prototype of its self-driving car. The company expects the vehicle to be hitting the streets of California some time in 2015. An early prototype of the self-driving car was shown off in May this year. In place of manual controls, such as a steering wheel, the car featured mapping systems and sensor technology that allowed it to "see" what was going on around it. There were, however, a number of features yet to be added, such as headlights. Since then, Google says it has been working with a number of different prototype vehicles, each being used to test the different systems required by an autonomous vehicle. Some of these have included the systems required to recognize road signs and road user gestures that are used on the city streets. These systems have now all been brought together in a first complete prototype. Although few specifications about the car have been released, it appears to sport a remodeled roof-top sensor module and newly-added indicators and headlights, as well as a lighter gray paint job. Google says its will continue testing the vehicles with its safety drivers for a "while longer." Source: Google

Self-driving car given UK test run at Oxford University

 

By Dave Lee Technology reporter, BBC News

A car that is able to drive itself on familiar routes has been shown off at an event at Oxford University.

The technology uses lasers and small cameras to memorise regular journeys like the commute or the school run.

The engineers and researchers behind the project are aiming to produce a low-cost system that "takes the strain" off drivers.

Other companies, such as Google, have also been testing driverless vehicle technology.

The search giant has pushed for law changes in California to allow its car to be tried out in real-life situations.

The Oxford RobotCar UK project is seeking to do the same in the UK, said Prof Paul Newman from Oxford University's department of engineering science.

"We're working with the Department of Transport to get some miles on the road in the UK," said Prof Newman, who is working alongside machine learning specialist Dr Ingmar Posner.

Gaining 'experiences'

Until the car can hit the streets, the team is testing it out in a specially-made environment at Begbroke Science Park in Oxfordshire.

Continue reading the main story

Analysis

Richard Westcott BBC transport correspondent at Begbroke Science Park in Oxfordshire

---

Frankly, it is a bit disconcerting being driven around by a robotic chauffeur, but then I remember thinking the same thing when I first used cruise control on a motorway.

It's amazing how quickly you adjust to things. Within five minutes I'd got used to the wheel turning on its own, and I wasn't remotely concerned when someone walked out in front of us (it was a tightly controlled safety experiment before anyone emails in, and the car did stop in plenty of time).

Fully autonomous cars won't appear in showrooms overnight. But it seems inevitable we will be handing over more of the driving to computers as the years roll by, and this Oxford University system could well be the next step.

There are barriers of course. Makers will have to prove they are safe. Then they'll have to convince the public. And there's the sticky question of who's liable if there's a crash.

Still, most car crashes are down to the human at the wheel. Plenty of people believe robotic cars could save thousands of lives in the future.

"It's not like a racetrack - it's a light industrial site with roads and road markings," Prof Newman told the BBC.

"Crucial for us, it can show our navigation and control system working.

"It's not depending on GPS, digging up the roads or anything like that - it's just the vehicles knowing where they are because they recognise their surroundings."

The technology allows the car to "take over" when driving on routes it has already travelled.

"The key word for us is that the car gains 'experiences'," Prof Newman explained.

"The car is driven by a human, and it builds a 3D model of its environment."

When it goes on the same journey again, an iPad built into the dashboard gives a prompt to the driver - offering to let the computer "take the wheel".

"Touching the screen then switches to 'auto drive' where the robotic system takes over, Prof Newman added.

"At any time, a tap on the brake pedal will return control to the human driver."

Spinning lasers

At the moment, the complete system costs around £5,000 - but Prof Newman hopes that future models will bring the price of the technology down to as low as £100.

Autonomous technology is being tested by several car manufacturers and technology companies.

Simple self-driving tasks, such as cars that can park themselves, are already in use across the industry.

The Holy Grail is a fully-autonomous vehicle that is location-aware, safe and affordable.

The iPad display tells the driver when the car is able to take over

Google has been testing its car for several years, with the company boasting of 300,000 computer-driven miles without an accident.

While at an earlier stage of development, Oxford University's car has significant key differences to Google's offering, Prof Newman said.

"Well if you look at it, we don't need a 3D laser spinning on the roof that's really expensive - so that's one thing straight away. I think our car has a lower profile."

He added: "Our approach is made possible because of advances in 3D laser mapping that enable an affordable car-based robotic system to rapidly build up a detailed picture of its surroundings.

"Because our cities don't change very quickly, robotic vehicles will know and look out for familiar structures as they pass by so that they can ask a human driver 'I know this route, do you want me to drive?'"

Prof Newman applauded Google's efforts in innovating in the space - but was buoyant about the role British expertise could have in the industry.

"This is all UK intellectual property, getting into the [driverless car] race.

"I would be astounded if we don't see this kind of technology in cars within 15 years. That is going to be huge."

 

Hidden Obstacles for Google’s Self-Driving Cars

 

 

Impressive progress hides major limitations of Google’s quest for automated driving.

• By Lee Gomes on August 28, 2014

Why It Matters

More than 1.24 million people die worldwide as a result of road traffic accidents each year, according to the World Health Organization.

Watch out: Google’s self-driving car can “see” moving objects like other cars in real time. But only a pre-made map lets it know about the presence of certain stationary objects, like traffic lights.

Would you buy a self-driving car that couldn’t drive itself in 99 percent of the country? Or that knew nearly nothing about parking, couldn’t be taken out in snow or heavy rain, and would drive straight over a gaping pothole?

If your answer is yes, then check out the Google Self-Driving Car, model year 2014.

Of course, Google isn’t yet selling its now-famous robotic vehicle and has said that its technology will be thoroughly tested before it ever does. But the car clearly isn’t ready yet, as evidenced by the list of things it can’t currently do—volunteered by Chris Urmson, director of the Google car team.

Google’s cars have safely driven more than 700,000 miles. As a result, “the public seems to think that all of the technology issues are solved,” says Steven Shladover, a researcher at the University of California, Berkeley’s Institute of Transportation Studies. “But that is simply not the case.”

No one knows that better than Urmson. But he says he is optimistic about tackling outstanding challenges and that it’s “going to happen more quickly than many people think.”

Google often leaves the impression that, as a Google executive once wrote, the cars can “drive anywhere a car can legally drive.” However, that’s true only if intricate preparations have been made beforehand, with the car’s exact route, including driveways, extensively mapped. Data from multiple passes by a special sensor vehicle must later be pored over, meter by meter, by both computers and humans. It’s vastly more effort than what’s needed for Google Maps.

Google’s cars are better at handling some mapping omissions than others. If a new stop light appeared overnight, for example, the car wouldn’t know to obey it. However the car would slow down or stop if its on-board sensors detected any traffic or obstacles in its path.

Google’s cars can detect and respond to stop signs that aren’t on its map, a feature that was introduced to deal with temporary signs used at construction sites. But in a complex situation like at an unmapped four-way stop the car might fall back to slow, extra cautious driving to avoid making a mistake. Google says that its cars can identify almost all unmapped stop signs, and would remain safe if they miss a sign because the vehicles are always looking out for traffic, pedestrians and other obstacles.

Alberto Broggi, a professor studying autonomous driving at Italy’s Università di Parma, says he worries about how a map-dependent system like Google’s will respond if a route has seen changes.

Michael Wagner, a Carnegie Mellon robotics researcher studying the transition to autonomous driving, says it is important for Google to be open about what its cars can and cannot do. “This is a very early-stage technology, which makes asking these kinds of questions all the more justified.”

Maps have so far been prepared for only a few thousand miles of roadway, but achieving Google’s vision will require maintaining a constantly updating map of the nation’s millions of miles of roads and driveways. Urmson says Google’s researchers “don’t see any particular roadblocks” to accomplishing that. When a Google car sees a new permanent structure such as a light pole or sign that it wasn’t expecting it sends an alert and some data to a team at Google in charge of maintaining the map.

In May, Google announced that all its future cars would be totally driver-free, without even a steering wheel. It cited the difficulties in assuring that a standby human driver would always be ready to take over. The company says it will initially test the new cars with the added controls now required by states that allow testing. But winning approval to test, much less market, a totally robotic car “would be a tremendous leap,” says David Fierro, spokesman for the DMV in Nevada, where Google now runs tests.

Among other unsolved problems, Google has yet to drive in snow, and Urmson says safety concerns preclude testing during heavy rains. Nor has it tackled big, open parking lots or multilevel garages. The car’s video cameras detect the color of a traffic light; Urmson said his team is still working to prevent them from being blinded when the sun is directly behind a light. Despite progress handling road crews, “I could construct a construction zone that could befuddle the car,” Urmson says.

Pedestrians are detected simply as moving, column-shaped blurs of pixels—meaning, Urmson agrees, that the car wouldn’t be able to spot a police officer at the side of the road frantically waving for traffic to stop.

The car’s sensors can’t tell if a road obstacle is a rock or a crumpled piece of paper, so the car will try to drive around either. Urmson also says the car can’t detect potholes or spot an uncovered manhole if it isn’t coned off.

Urmson says these sorts of questions might be unresolved simply because engineers haven’t yet gotten to them.

But researchers say the unsolved problems will become increasingly difficult. For example, John Leonard, an MIT expert on autonomous driving, says he wonders about scenarios that may be beyond the capabilities of current sensors, such as making a left turn into a high-speed stream of oncoming traffic.

Challenges notwithstanding, Urmson wants his cars to be ready by the time his 11-year-old son is 16, the legal driving age in California. “It’s my personal deadline,” he says.

This story has been updated by editors after clarifying information from Google.