Tag Archives: driverless cars

Primitive Driverless Ford Cars Scheduled For 2017

Last April, Sebastian Thrun explained why we need driverless cars in a great TED talk, but even though it’s clear that automated driving is a lot safer and more convenient, it’s also clear that social stigmas and legal issues will require automation to happen incrementally, in probably three phases:

In a few years, we might see some enhanced version of cruise control that actually controls the car during cruise conditions on the highway. Then during traffic, and finally in the city. All requiring an alert driver of course, to shoulder the blame if anything should go wrong.

About a month ago, a version of that first phase was successfully tested: a self-driving caravan of cars — a “road train” that roams the highways, which cars can join and leave at will, and which controls the car as long as it’s virtually attached to the caravan. And last week, Ford announced that in five years, the second phase will come to fruition, via a technology they’re calling Traffic Jam Assist.

The idea is that the car will drive by itself in traffic jams on highways — at low speeds on limited-access roads with well-marked lanes, and devoid of people and animals. It will accomplish this with sensors that self-parking cars already have on board: cameras, sonar and radar. They enable behaviors like adaptive cruise control — in which the car can slow down and speed up based on the car in front of it — and lane-keep assistance that prevents the car from drifting out of its lane. Put these together, and the car can drive itself until you have to change lanes.

 

The highway limitation is there no doubt to avoid legal issues: if a bicyclist gets hit by a driver who claims the car was driving, even if the driver is proven wrong, the PR damage to Ford will already be done — much as what happened to Toyota amid the “unintended acceleration” scandal of 2009. But after a period of public adjustment to the idea — and some off-highway use by more risk-tolerant drivers — the highway limitation will probably be dropped, followed in a few years by a new feature which will enable the car to change lanes and make turns. If Ford sticks to its timeline for this second phase, it seems plausible that third-phase, fully autonomous vehicles will be available by 2025. Although, the driver will likely still have to technically be “in control” of the car.

From Extreme Tech, via Slashdot

See also:

 

Self-Driving Caravan Test A Success

While in individualistic America the autonomous driving push has been behind standalone cars, in Europe the first successful test was that of a caravan known as Project SARTRE (Safe Road Trains for the Environment). The concept is also known as a road train: cars attach themselves (virtually, not physically) to a road train that guides them along, and when they’re ready to get off, they veer off and the road train re-configures itself. While in the train, each car follows the one in front of it, and directions on what to do are passed from car to car via a wireless network that the train creates on the fly. The technology behind it is built by Volvo and has been tested on private courses, but this test was done on 125 miles of public roads in Spain, traveling at about 50mph.

See also:

From Reg Hardware, via Slashdot

The Driverless Car Leading The Blind

For the past couple of years, Google has been leading the effort to create fully automated cars that drive themselves. The benefits you generally hear being touted about are related to safety and convenience:

  • car accidents are generally caused by human error
  • robots are always vigilant, while humans get easily bored when driving and are thus prone to distraction by anything at all: gadgets, eating, talking, sleeping, etc;
  • proper, robotic driving would alleviate traffic
  • your self-driving car is also your taxi and valet; plus you’ll never get a speeding ticket again
  • you could go to sleep in your car in Miami and wake up in Orlando

 

Blind man sitting in the driver's seat of Google's self-driving Prius

 

Well Google decided that wasn’t enough, and that it was time to also play up the heart-warming angle: driverless cars can also give independence to people who can’t drive. The technology isn’t quite ready for the public yet, so they shot a demo video of what things could be like, probably somewhere around 2020: a blind guy gets in his car, tells it where to go, and it takes him through the drive-thru at Taco Bell, then to pick up his dry cleaning, and on the way back home he eats his burrito.

Where this would change my life is to give me the independence and the flexibility to go to the places I both want to go and need to go, when I need to do those things. — Steve Mahan, blind test driver

What the video implies is that shut-ins would no longer have to spend all day locked in their house, be at the mercy of their friends and family, or have to resort to expensive taxis.  Besides the blind, this includes all kinds of disabled people: many of the elderly, the deaf, some of the physically handicapped, children. Wait …. children?! Ugh. Back to the drawing board…

 

See also:

 

From Fox News, via Slashdot

Driverless Cars Now Legal For Test Driving In Nevada

Back in June of 2011, the Nevada legislature, at Google’s behest, told their DMV to come up with rules to allow driverless cars to roam the highways. Now those rules are finally in place and beginning March 1st, companies can apply for robotic driver licenses — for testing purposes only. The robotic test cars will have red license plates, but the Nevada DMV also prepared for a future when robotic cars will be sold to the public: general-use robotic cars will have neon-green license plates.

 

Autonomous Prius at the 2011 Tokyo Motor Show

 

In the past few months, three more states have been considering laws that would follow Nevada’s lead: Florida is poised to be next, followed by Hawaii and Oklahoma.

See also:

From NPR and The Orlando Sentinel

Bans On Driving While On Cellphone Reduce Cellphone Use, But Not Accidents

One of the bigger news stories this week was the NTSB’s recommendation that state governments should ban all use of personal electronic devices (cellphones, iPods, etc) while driving, even ones with hands-free capability. The theory is that when you’re on the phone, your mind is busy with the conversation, and so you’re distracted even if your hands are free. The natural response to this is to ask if passengers and babies should be banned from cars too, but they say talking to a passenger will still be tolerated: since they’re part of the shared experience, a passenger will stop talking in a dangerous situation and maybe even alert the driver to potential hazards. But of course, talking on the phone with it in your hand is ludicrously dangerous, and texting! — you might as well just be the devil for doing something that menacing. The data, however, would beg to differ.

The accident behind the NTSB's electronics ban recommendation

 

The NTSB cites the cause of its recommendation as an accident from August 2010 in which a school bus ran into another school bus that had crashed into a pickup truck that had crashed into a semi that had slowed down due to construction; the pickup truck driver was killed, as was a high school girl on the first bus; dozens were also injured. The pickup driver had sent and received 11 texts in 11 minutes prior to the accident, with the last one having come in just before he crashed, so what happened is clear: he got the text, looked at it, missed the semi slowing down, and crashed into it. Ergo, we should ban all electronic devices.

But then again, why were the two buses following so close behind the pickup truck that they couldn’t have stopped in time or avoided it? The NTSB has been teaching the two-second rule for years, to prevent these types of accidents. And even if the pile-up was all the pickup driver’s fault, isn’t it a bit of an overreaction to ban all electronics due to one accident? It’s reminiscent of grade school, when the teacher is writing on the board with her back to the class, and because Billy throws a wad of paper at her, everyone gets punished. Or when one idiot tries to blow up a plane with bombs in his shoes, and then billions of people have to take their shoes off at the airport.

 

“But wait,” you say, “it’s not a punishment — it actually is dangerous to text and drive, so it’s a safety measure, for our own good.” That’s true, and texting is not alone; there are interminable examples of distracted or impaired driving: driving while eating, driving while the baby’s crying, driving while old, driving while sleepy, driving while putting on makeup, driving while tuning the radio, etc, etc, ad nauseam. Given how hard it is to enforce bans on the vast majority of these activities, and how creative people are at coming up with more, is it really worthwhile to try to specifically ban every one of them?

The auto insurance industry, which has a clear financial interest in reducing accidents, set up an Institute for Highway Safety in 1959. They released a report (PDF) in September of 2010 on the effects of driving while texting (DWT). It found that crashes actually went up slightly in states that had enacted a ban on DWT, possibly because it’s even more dangerous to drive while hiding the fact that you’re texting. It also said that “texting bans failed to produce a detectable reduction in crash risk,” and bans on specific driver distractions do not work: drivers will always find some novel distraction — even in a world without electronics — because, and this observation wasn’t in the report, the vast majority of the time, driving is really, really boring. (More highlights from the report are reproduced further below.)

The report also mentions that somehow, even though texting while driving is obviously dangerous, and texting has increased substantially in the past decade — about 18% of all drivers text while driving; almost 50% of college-age drivers, — car accidents have not increased to keep up with this new menace. In fact, the institute told NPR that “…with more and more people having phones in their cars and using them, the number of overall crashes has been declining.”

So it’s clear that the NTSB’s recommendation is sorely misguided, ineffective, and simply an overreaction looking for a scapegoat to a major accident. But the problem they’re trying to address is real: 40,000 people die every year in car accidents, and as the institute’s report said, 90% of them are due to driver error. That error will not go away as long as people keep driving cars. Therefore, what the NTSB should really recommend is a ban on driving, the minute Volkswagen, GM and Toyota come out with driverless cars.

If you haven’t seen Sebastian Thrun’s TED Talk on Google’s driverless car, it’s a good way to spend 20 minutes:

 

Some highlights of the conclusion from the insurance industry’s report:

Insurance collision loss experience does not indicate a decline in crash risk when texting laws are enacted. Rather, there appears to have been a small increase in claims in the states enacting texting bans, compared to neighboring states.

….

This unexpected consequence of banning texting suggests that texting drivers have responded to the law, perhaps by attempting to avoid fines by hiding their phones from view. If this causes them to take their eyes off the road more than before the ban, then the bans may make texting more dangerous rather than eliminating it.

….

The results of this study seem clear. In none of the four states where texting bans could be studied was there a reduction in crashes. […] If the goal of texting and cellphone bans is the reduction of crash risk, then the bans have so far been ineffective. Bans on handheld cellphone use by drivers have had no effect on crashes (HLDI, 2009), as measured by collision claim frequencies, and texting bans may actually have increased crashes.

….

In four states, texting bans failed to produce a detectable reduction in crash risk, despite the geographic dispersion of these states and their controls. It is unlikely that uncontrolled covariates are confounding all the results. This is similar to the previous study, showing that hand-held cellphone bans have not affected crash risk in four different states.

….

These results indicate that distracted driving crashes are a complicated issue unlikely to be affected greatly by laws banning only one or another potential distraction. Distracted driving has long been a major contributor to the motor vehicle crash problem in the United States. In 1979, a report on the “Indiana Tri-Level Study” concluded that “driver error” had been the proximate cause of 9 out of 10 crashes investigated

….

The long history and ubiquity of distracted driving crashes, coupled with the current findings, suggests that public policy that focuses on only one source of distraction (for example, cellphone conversations or texting) may fail simply because it doesn’t recognize that drivers always are subject to distraction. Taking away cellphones may result only in drivers defaulting — even unintentionally — to new (or old) forms of distraction.

 

From The Institute of Highway Safety (PDF), via NPR

Toyota Is Making A Driverless Prius

Seems like thanks to Google and DARPA, this driverless car thing is really taking off. At the 2011 Tokyo Motor Show, Toyota is letting people test-ride in an autonomous version of the Prius called the AVOS (Autonomous Vehicle Operation System). The passengers can summon the car, get in the back seat and experience the AVOS avoiding obstacles and parking itself. A Gizmag article also points out that since they both sponsored winning teams in DARPA’s autonomous vehicle Grand Challenges, GM and Volkswagen are generally thought to be at the forefront of the driverless car game, which is probably making Toyota nervous.

 

Autonomous Prius at the 2011 Tokyo Motor Show

 

 

All of this is great, because thanks to human drivers, many more Americans have died in car accidents than in all the wars we’ve ever fought, put together.

Also check out Why We Need Driverless Cars.

From Gizmag, via Slashdot

Computers Took Our Middle-Class Jobs

The peak of the American economy before the recent recession happened at the end of 2007, when the GDP was 13.33 trillion$ and 138 million people were employed. After almost 4 years (three times longer than our average recovery time), the GDP is finally back above that level, to 13.35 M$ — but with 7 million fewer workers. Those people have been replaced partly by their former, more productive co-workers who picked up the slack, partly by cheaper workers in India, China and the like, and partly by computers who have gotten a lot more adept at doing sophisticated things around the office.

For example, there’s a lot less need for secretaries these days: between Outlook, GMail, the decline of postal mail, easy-to-use calendars with reminders on all the smartphones, files becoming electronic without need for filing (thanks to Google desktop search), and now even Siri, a lot of a secretary’s job has been largely replaced by software from Google and Apple. Travel agents have been replaced by Travelocity and Expedia (dot cooom). Accountants, by Excel, Turbotax, and Mint. Broadcast engineers, by streaming Internet video. Police and security officers, by cameras and security systems. Paralegals, by scanners, optical recognition, text search software, and the Internet. Librarians, by Google Books and Amazon. Bank tellers, by ATMs.  Even bakers, by vending machines. All middle class jobs that are disappearing.

And that’s just the beginning: Watson — the IBM computer that mopped the floor at Jeopardy! with the two best people that ever played the game — is going to replace all kinds of Jeopardy! contestant-like jobs that depend on recalling trivia from large amounts of knowledge: doctors, lawyers, government employees, etc. Of course, technology doesn’t eliminate those jobs completely: we’ll need surgeons, trial lawyers and DMV clerks for the foreseeable future. Same goes for the other jobs (e.g., we’ll always need some accountants and traffic cops), because we’re unlikely to ever automate everything, but what will happen is that one person will be able to do the job of several. Agriculture was one of the first industries to become automated, and now a couple of people can run an entire giant farm that used to take dozens; over the past century, those jobs went from 38% to 2% of the workforce.

 

Watson on Jeopardy!

 

As a specific example of what’s coming, NPR has an article that highlights how disruptive automation will be to the auto industry: in the past 30 years more Americans have been killed by cars than by all the wars in the past 300 years — 1.5 million people. Because of that, and also because they’re really cool, driverless cars are coming to improve safety and efficiency. When they do, right off the bat, they’ll put a lot of people out of work: taxi drivers, valets, bus drivers, traffic cops, etc. And people that don’t drive too often may forego owning a car altogether and just use services like Zipcar; this means less need for mechanics and parking maids, less government revenue from drivers license fees, registration fees, traffic tickets, parking tickets, etc.

 

Google's driverless car. Illustration by The Globe And Mail

 

One interesting thing is that lower-skill jobs are not yet being targeted by robots, because those jobs tend to pay too little and aren’t worth replacing by an expensive machine that does them probably much worse: all kinds of cleaning jobs (janitors and maids, garbage men, bus boys), fast food cooks, animal caretakers, etc. And then there are the jobs no one wants to see a machine doing, mostly in the service industry: waiters, receptionists, nannies, hair dressers, retail salesfolk. Because of the high-cost and largely impersonal nature of middle-class jobs, it’s mostly those that are at risk for being replaced by machines.

So what do you do when the metal ones decide to come for you? Besides getting Old Glory Insurance, re-education is pretty much the only alternative to hoping you’re one of the few that get to stay behind. The loss of jobs to automation is nothing new and has been an issue since the days of John Henry, almost 150 years ago. And so far, every time an industry is disrupted by machines, a new industry pops up to take its place and create the new jobs needed. For example, when agriculture started to decline, radio and television were invented, as were cars and airplanes. Now, at the very least, we’re going to need people to design all those new robots… until someone builds one that can reproduce.

And in case you’re interested in some of that robot insurance:

From NPR, NPR, Forbes and Bloomberg

Human Driver Crashes Google’s Driverless Car

As if to really drive home the point how much better robots are at driving cars than us, a Google car that can drive by itself rear ended another car which rear-ended another and that one rear-ended a fourth, all while it wasn’t in auto-driving mode: a real person was driving it. Skeptics will obviously wonder if it’s not a big cover-up and that the human wasn’t actually in control, but it’s hard to imagine that the car’s sensors failed to slow the car down enough to prevent a 4-car pile up, all while the human driver was watching. But it makes a lot of sense that the auto-pilot was off and the person driving wasn’t paying attention. Which is exactly why we need driverless cars.

The Google driverless car has that thing on its roof

 

From Jalopnik, via Slashdot

Driverless Cars Coming To Nevada

Good news for fans of cars that drive themselves: Nevada is now the first state that allows them on highways. The legislature directed their department of transportation to write up rules which would regulate the robotic drivers, which in turn should prevent most accidents, since they are generally caused by human error. Hopefully it won’t be long before you’ll be flagging down driverless taxis on the Las Vegas strip, as prophesied by Total Recall, more than 20 years ago.

Via The LA Times

Why We Need Driverless Cars

Sebastian Thrun gave a TED talk recently about why the driverless cars that came out of DARPA’s Grand Challenge and are being developed at Google aren’t just a nice feature like cruise control, but rather a necessity for society. The vast majority of car accidents are caused by human error, and they are the leading cause of death for young people. Robotic cars could vastly improve highway capacity and eliminate traffic jams. They would allow people to reclaim the average hour spent on commuting and do something useful with it. And the various forms of distracted driving — sleepy driving, drunk driving, texting while driving, shaving while driving — would be eliminated.

The technology for driverless cars is here: Thrun’s cars have successfully driven over 140,000 miles on all kinds of roads and conditions, from highways to dense city streets, including California’s Highway 1 and the “crookedest street in the world”, San Francisco’s Lombard Street. The EU is also developing a similar concept of the “road train,” made up of driverless cars. Right now all of this is at the prototype stage, but it could be brought to market in probably around a decade.

The big issue isn’t the technology, but rather the legality. If you get in a car accident now, you or the other driver are likely responsible. But if two driverless cars are involved in an accident, where does the blame go? To you? To the other car? To Google? And what if someone gets injured or killed? Would Google get sued for millions each time? No car manufacturer would take on that liability.

The truth is, as a society, we’d rather deal with the 10 million car accidents per year and the 39,000 deaths they cause because, at least humans were responsible. To put that into perspective, it would be as if everyone in New York City got in a car accident every year, and everyone in Greenwich Village died as a result. We don’t want machines killing us, even if the total number would be much, much lower. We’ve already had a sneak peek of the coming backlash, via the scandal surrounding “unintended acceleration” in Toyota cars. For over a year, the cars were blamed for a couple dozen deaths, until most of these turned out to be caused by driver error. It makes you wonder how many at-fault drivers, having heard of the acceleration problems, decided to blame their accident on the car? After all, that instinct is only human.

But we do want driverless cars after all, so what will probably happen is that cars will gradually become more and more “automatic”, until we get used to the idea and the legal ramifications get sorted out. Cruise control has been around for a couple of decades, and self-parking cars for a couple of years. “Lane-departure warning systems” which detect your car drifting out of its lane are becoming more common, as are collision warning systems which predict that you’re about to crash.

In a few years, we might see some enhanced version of cruise control that actually controls the car during cruise conditions on the highway. Then during traffic, and finally in the city. All requiring an alert driver of course, to shoulder the blame if anything should go wrong.