Google Lobbies Nevada To Allow Self-Driving Cars

[M II A II R II K]

Google Lobbies Nevada To Allow Self-Driving Cars


May 10, 2011

By John Markoff

Read More: http://www.nytimes.com/2011/05/11/science/11drive.html?_r=1&hp


Google, a pioneer of self-driving cars, is quietly lobbying for legislation that would make Nevada the first state where they could be legally operated on public roads. And yes, the proposed legislation would include an exemption from the ban on distracted driving to allow occupants to send text messages while sitting behind the wheel. The two bills, which have received little attention outside Nevada’s Capitol, are being introduced less than a year after the giant search engine company acknowledged that it was developing cars that could be safely driven without human intervention.

- Last year, in response to a reporter’s query about its then-secret research and development program, Google said it had test-driven robotic hybrid vehicles more than 140,000 miles on California roads — including Highway 1 between Los Angeles and San Francisco. More than 1,000 miles had been driven entirely autonomously at that point; one of the company’s engineers was testing some of the car’s autonomous features on his 50-mile commute from Berkeley to Google’s headquarters in Mountain View.

- Policy makers and regulators have warned that the technology is now advancing so quickly that it is in danger of outstripping existing law, some of which dates back to the era of horse-drawn carriages. New laws will be required, they argue, if autonomous vehicles are to become a reality. Policy analysts say Nevada is the first state to consider the commercial deployment of a generation of vehicles that may park themselves, perform automatic deliveries or even act as automated taxis on the Las Vegas casino strip.

- “What if I could take out my phone and say, ‘Zipcar, come here,’ †he asked an industry conference last year, “and a moment later the Zipcar came around the corner?†Google’s autonomous vehicle ambitions hint at an emerging vehicle-industrial complex in Silicon Valley. Mercedes, Volkswagen and other carmakers have laboratories in the region, I.B.M. has a battery development initiative, and the Nummi plant in Fremont, once a joint venture of General Motors and Toyota, has been reopened by Tesla.

.....

 

[afransen]

Robocars have some liability issues. One would assume the driver retains all liability for property damage, loss of life and limb as a result of automated operation of such a vehicle. The first manslaughter case will be interesting. Who is liable when there is no human passenger? These are pretty far from trivial legal concerns.

Besides that, robocars don't really address capacity issues associated with car usage. They ought to be able to make highways flow better by preventing asshats from passing/cutting people off, but that flow rate will be restricted to one vehicle every 3 seconds in each lane or so in good weather, and more in poor weather. We'll still need signalized intersections to allow traffic to cross and allow pedestrians to cross streets. So we might be able to squeeze another 30% capacity out of the network on a good day. Bad days will be as bad as they are now. Robocars will be a useful technology when mature, but I don't think they'll represent a paradigm shift. PRT can more likely play that role, if anything.

 

[Mapleson]

Besides that, robocars don't really address capacity issues associated with car usage. They ought to be able to make highways flow better by preventing asshats from passing/cutting people off, but that flow rate will be restricted to one vehicle every 3 seconds in each lane or so in good weather, and more in poor weather. We'll still need signalized intersections to allow traffic to cross and allow pedestrians to cross streets. So we might be able to squeeze another 30% capacity out of the network on a good day. Bad days will be as bad as they are now. Robocars will be a useful technology when mature, but I don't think they'll represent a paradigm shift. PRT can more likely play that role, if anything.

Robocars would not have the same inherit limitations at human-intervention vehicles. Lane capacity for normal vehicles is 1200 per hour (1 car per 3 second). The two main components a robocar beats out old tech are dectection and reaction. By reducing the time needed between observation and correction, you can reduce headways. If you add the not-too-far-fetched concept of car communication networks, you could have vehicles accelerate and break in tandem.

Intersections will still be an issue, but if the first 5 cars at the lights started moving together, you could see larger traffic blocking. We'll never build our way out of traffic problems, but smart technology allows the best use out of diminishing assets.

The bigger note I'd make is that Nevada is home to U-Haul due to their non-renewal permanent licence plates. I wonder if Google is planning robotaxis...

 

[M II A II R II K]

How Google's Robot Cars Will Revive Sprawl


Read More: http://www.fastcompany.com/1758128/...iquitous-and-commuting-less-painful-revive-sp


Testifying before the Nevada State Assembly in April to have the state legalize its driverless cars, Google lobbyist David Goldwater asked the state’s transportation committee to “imagine a time when we will be able to call our public transportation on our cell phones or smart phones and tell it to come to our door to pick us up, without anybody in it, take us to our job, and be released to go perform the same service for somebody else.” But a world filled with robot cars may have consequence even their creator can’t predict. Driverless cars would be a perfect match for car-sharing services such as Zipcar or Getaround, gradually replacing the idea of car ownership with “mobility-as-a-service.”

- That, in turn, may lead to a precipitous fall in car ownership--as high as 50%--while breathing new life into suburbia and creating more congestions as the pain of commuting lessens. And what would halving the number of cars on the road mean for the Detroit Three--and the taxpayers who’d like the rest of their bailout back? Six months before that hearing, when they were first reported in The New York Times, Google’s test cars had traveled 140,000 miles with only one accident. (A car was rear-ended while stopped at a light.) Since then, its car czar Sebastian Thrun--a former Stanford AI professor who won DARPA’s autonomous vehicle competition in 2005--has been vocal about the possibilities of cars without drivers, including fuel efficiency and fewer fatalities. And, perhaps most importantly, a future without horrible commutes.

- But what if driverless cars mimicked transit instead? What if you could watch movies, make Skype calls, or play with your children in the back seat until you were delivered to your doorstep? A game-changer in terms of commuting time and cost, it would breathe new life back into suburban sprawl; it's not a problem to live two hours from work when you can spend those two hours as if you were on your couch. And what if you didn't even have to be making payments on that car that was driving you to work. Robot cars could allow us to drastically reduce car ownership.

- “Many of today’s young people living in a more urban society learn to live without cars,” theorized the environmentalist Lester R. Brown at the time. “They socialize on the Internet and on smart phones, not in cars.” Perhaps in the future, they will socialize on smart phones in driverless cars. In that case, to survive, the auto industry's business model might be starting to look like the smart phone one--with the hardware made in China, and the brains supplied by Google.

.....

 

[afransen]

Robocars would not have the same inherit limitations at human-intervention vehicles. Lane capacity for normal vehicles is 1200 per hour (1 car per 3 second). The two main components a robocar beats out old tech are dectection and reaction. By reducing the time needed between observation and correction, you can reduce headways. If you add the not-too-far-fetched concept of car communication networks, you could have vehicles accelerate and break in tandem.

Intersections will still be an issue, but if the first 5 cars at the lights started moving together, you could see larger traffic blocking. We'll never build our way out of traffic problems, but smart technology allows the best use out of diminishing assets.

The bigger note I'd make is that Nevada is home to U-Haul due to their non-renewal permanent licence plates. I wonder if Google is planning robotaxis...

You need a fairly tightly controlled fleet for lower headways to be anything much less than three. If you allow headways under three seconds, I think PRT makes more sense, since grade separated vehicles don't have to stop and won't get flattened by errant trucks.

 

[kEiThZ]

People are getting too worked up about this. Google is pushing this so that they can develop the tech to save money on their Streetview service.

That and Google just loves to push the boundaries of tech. Look at some of the things they've done for solar power.

 

[Mapleson]

You need a fairly tightly controlled fleet for lower headways to be anything much less than three. If you allow headways under three seconds, I think PRT makes more sense, since grade separated vehicles don't have to stop and won't get flattened by errant trucks.

Why wouldn't you use robotrucks first? Reduce shipping costs would be a huge competive edge.

Where do you think the 3-second minimum headway rule comes from? Generally, one second for human observation, half a second for human response, and 1.5 seconds for deceleration. Only in a human-human interaction would a 3-second headway need to be maintained. Robot-human interactions would reduce the first 1.5 seconds by a factor of 3 or more. Robot-robot interactions would allow for the second 1.5 seconds to be reduced as well, with the trailing vehicle better able to maximize deceleration per time.

The underlying assumptions are valid, until you have a game-changer, like this technology. It just needs time for people to accept.

 

[afransen]

3 s is brickwall stop criteria. It's insane, I agree, especially in a grade-separated computer controlled environment.

On the other hand, there are a multitude of things that can go wrong on a highway in mixed traffic with trucks and human-controlled vehicles. For liability and political/regulatory reasons, I can't imagine sub 3 second headways being permitted for a good long time. That's okay. Robocars would still be very useful without that. But they won't be making all our road capacity problems disappear any time soon. If anything, the reduced time/effort cost of driving will increase vehicle-kms driven and probably result in no net decrease in congestion.

 

[Jonny5]

The Nevada legislature has approved this.

 

[Mapleson]

3 s is brickwall stop criteria. It's insane, I agree, especially in a grade-separated computer controlled environment.

On the other hand, there are a multitude of things that can go wrong on a highway in mixed traffic with trucks and human-controlled vehicles. For liability and political/regulatory reasons, I can't imagine sub 3 second headways being permitted for a good long time. That's okay. Robocars would still be very useful without that. But they won't be making all our road capacity problems disappear any time soon. If anything, the reduced time/effort cost of driving will increase vehicle-kms driven and probably result in no net decrease in congestion.

Agreed. It'll be 10-20 years before we see any profound shifts in demographics, but if this catches on well in Nedava, then like seatbelts or daytime running lights, I see it becoming mandatory for at least main arterials.

Some interesting wording in the Nedava law...

The Department shall by regulation establish a driver’s license endorsement for the operation of an autonomous vehicle on the highways of this State. The driver’s license endorsement described in this subsection must, in its restrictions or lack thereof, recognize the fact that a person is not required to actively drive an autonomous vehicle.

They're restricting who can drive an autonomous can, at least for now.

...may exercise the privilege thereby granted upon all streets and highways of this State and shall not be required to obtain any other license to
exercise such privilege by any county, municipal or local board or body having authority to adopt local police regulations.

This is key to stopping NIMBYism when they set-up the testing areas.

“Qualified alternative fuel” means compressed natural gas, hydrogen or propane.

At the same time they are opening up alternative fuel sources and making local authories provide free parking ($10 per year max charge for a decal).

“Qualified alternative fuel vehicle” means a motor vehicle that:
1. Is equipped with four wheels;
2. Is made by: (a) An original equipment manufacturer; or (b) A qualified vehicle modifier of alternative fuel vehicles;
3. Is manufactured primarily for use on public streets, roads and highways;
4. Has a manufacturer’s gross vehicle weight rating of less than 8,500 pounds;
5. Can maintain a maximum rate of speed of at least 70 miles per hour; and
6. Is propelled:
(a) To a significant extent by an electric motor which draws electricity from a battery that:
(1) Has a capacity of not less than 4 kilowatt hours; and
(2) Can be recharged from a source of electricity that is external to the vehicle; or
(b) Solely by a qualified alternative fuel, and meets or exceeds the federal Tier 2 bin 2 exhaust emission standard, as set forth in 40 C.F.R. § 86.1811-04.

I find it funny they are willing to let a car drive itself, but maintain it must have 4 wheels.

 

[TrickyRicky]

Self-driving cars equals nightmare scenerio. It's not the technology it's the concept. The technology and conceptualization for the technology has existed for decades. It's just a bad idea. The only way I would start to be OK with this is if the vehicles ran on a private for profit separated highway and road infrastructure running vehicles that were not privately owned by individual people.

 

[Mapleson]

It's just a bad idea.

Any reason why, or just because you said so? Do you believe human drivers in privately owned vehicles is the better idea?

[video=youtube;t9Fxp3HK6DI]http://www.youtube.com/watch?v=t9Fxp3HK6DI[/video]

Some ping-pong robots to show more precise high-speed reactions:

Topsy the Ping Pong Dominator
Two balls, one paddle

 

[TrickyRicky]

Yes, of course it being a bad idea is my opinion. It's like using hydrogen for fuel. These technologies have been around for a long time. They were demonstrating caravaning self-driving cars over a decade ago. I even attended an ITS (intelligent transportation systems) show myself over a decade ago. It's not the technology it's the concept.

A self driving car will certainly be able to beat the preformance of a human driver under ideal conditions. The key is ideal conditions. Conditions need to be ideal based both on the externalities of the road (conditions, unexpected events, etc.) and internally (the state of repair and functionality of the vehicle and its guidance systems).

The trouble is in the real world external and internal conditions are not ideal. Guidance systems will be useful in the future to assist or augment or amplify the response of the human. I have no problem with this. However, who makes sure that the external and internal conditions remain ideal for the automated self-driving car?

Please keep in mind the engineering difference between say the brakes of the car wearing out and the guidance system of an automated car wearing down or being mis-aligned. On the one hand the mechanical breaking preformance of the car erodes over time. You could probably drive your car functionally for years with substandard breaks. On the other hand the guidance system of an automatic car wears down, or becomes mis-aligned...bam everyone is dead instantly. How then can we allow individuals to purchase or maintain their own vehicles or have such vehicles travel on public mixed-traffic roadways when they are constantly 1 second away from causing an accident?

 

[Mapleson]

So Google's AutoCar driving for 225,000 km as of last October in real world conditions with one accident (a rear-ending by a human while stopped at lights), doesn't mitigate your catastrophic visions? The cars do have a manual override mode. I can think of two additional fail safes that would be ample protection from deterioration: an annunal/bi-annual service inspection like we have for vehicle emissions; and, a self-diagnostic check that prevents the car from driving itself if it's less than 99.99% perfect.

What about self-driving trains? If the guidance system 'wears down', the train might not break for an upcoming curve and 'bam everyone is dead instantly'. We don't worry about this because there is sufficent fail-safes in the system. What about autopilot on planes? If the guidance system 'wears down', the plane might nosedive into the ground and 'bam everyone is dead instantly'. What about robot surgeons? If the guidance system 'wears down', the scapal might slip and 'bam someone is dead instantly'. What about lifeguard robots? If the guidance system 'wears down', it could take you out to sea.

The question is even if such a catastrophic failure occured, would it happen more or less frequently than human judgement leads to a road fatality. For Canada in 2009, there were 2209 people killed in 2011 collisions; 1 fatality per 150 million vehicle-kilometres or 1 per 10,526 drivers per year. We need a larger sample size for a valid statistical comparison, but so far in real-world non-ideal conditions, Google's Autonomous Car is flawless.

 

[M II A II R II K]

Only Five Networked Cars For Every 1,000 Would End Traffic


June 28th, 2011

By Matthew Battles

Read More: http://www.fastcompany.com/1763705/networks-reduce-traffic-jams-even-when-theyre-full-of-holes

Additional: http://telematicsnews.info/2011/06/26/eu-opel-presents-findings-of-telematics-v2x-field-test_jn2262/


It’s safe to say, despite Nevada's recent approval of fully automated vehicles, that we Americans are ambivalent about the prospect of networked cars. Despite the obvious safety and convenience we would glean from a fleet of autos that could negotiate traffic autonomously, avoid pedestrians and potholes, and park themselves, the myth of the independent driver is a powerful one in our culture. Fortunately, the wired automobile is not an all or nothing affair; researchers afiliated with Opel reported last week that it takes as few as five wired cars in every 1,000 to sketch an accurate picture of traffic conditions that engineers can use to respond to tie-ups and reduce congestion.

The project, called Diamant (Dynamic Information and Application for Mobility with Adaptive Networks and Telematics Infrastructure) consists of automobile-mounted, Wi-Fi-enabled sensors, which relay traffic data from car to car until they reach a roadside base station that sends the info to a control center, where engineers can monitor traffic jams, accidents, and construction zones and mount responses in the form of radio alerts and text messages. The surprising discovery is that even when such an automotive web is loosely knit and full of holes, connecting as little as .005 percent of cars on the road, the information it provides can help traffic managers ease congestion, potentially saving hundreds of millions of dollars in fuel costs--not to mention reducing the stress and anxiety of drivers, whether their rides are Wi-Fi-enabled or not.

.....