backtop


Print 109 comment(s) - last by lagomorpha.. on Jun 4 at 9:11 AM


  (Source: Getty Images)
GM is taking the project seriously

General Motors Corp. (GM) gave Google Inc. (GOOG) its begrudging regards this week after hearing about the Mountain View, Calif. company's vision for autonomous vehicles.
 
I. GM on Google -- It's a "Real... Threat"
 
While it looked a bit like a Volkswagen AG (ETR:VOW) Beetle smacked with the ugly stick a few more times, Google's in-house design for an autonomous electric vehicle (EV) is turning heads based on the sophistication of its driving and collision avoidance.  Google is building a fleet of 100 of the bulbous 2-seaters to test its smart driving technology.
 
At an event in the Detroit, Mich. area, GM's Product Development chief, Mark Reuss was inevitably asked to weigh in on the excitement surrounding the Google fully autonomous smart car.  He did not hold back, responding:

Anybody can do anything with enough time and money.  If they set their mind to it, I have no doubt [that they will be] a very serious competitive threat.  [The car is] kind of cool [and looks sort of like a VW Beetle].

[Automation is] going to be a creep, it’s not going to be a mind-bending thing.  I don’t think you’re going to see an autonomous vehicle take over the city anytime soon
.

Google automated car
Google's automated car is a "threat" according to GM's Product Development chief.

GM certainly seems an authoritative voice on the topic.  
 
In April 2014 in the U.S. it sold over 254,000 automobiles internationally, ahead of Ford Motor Comp. (F) (210,000+) and Toyota Motor Corp. (TYO:7203) (199,000+) [source].  GM was sales king of the auto industry longer than any other company in the history of the automobile.  
 
From 1931 to 2007, GM sold more cars and trucks than any other automaker.  And even as it's struggled to hold back a surging Ford and to regain its lead from Toyota, GM is still seeing strong sales.  (Volkswagen was the only automaker to beat it in 2013, a bit of a surprise.)
 
And it's good to see such straight talk from GM and an earnest assessment of its possible future competitor.
 
II. The Race to a Smart Car
 
For years we heard lots of automakers pay lip service to the concept of autonomous driving, with some even spending a good deal of money and time investigating it.  GM was perhaps foremost in the field prior to its bankruptcy.  Over a half-decade ago, back in Jan. 2008 we road along in a modified 2008 GM Tahoe which won the DARPA's (Defense Advanced Research Projects Agency) 2007 Urban Challenge.

GM Darpa
GM's 2008 Chevy Tahoe, modified for the 2007 DARPA Urban Challenge
[Image Source: Jason Mick/DailyTech LLC]

As the car zipped around the obstacle course in sunny Las Vegas, Nev. it quickly became apparent why the vehicle won the smart car challenge -- it was pretty good at avoid collisions... really good, in fact.  As a demo, a second test driver would cut off the GM car or otherwise block its path.  In every case the GM vehicle knew what to do, performing better than many human drivers would in such a case.
 
But after the bankruptcy rolled around in mid-2009, the pace of development has slowed, in some regards, allowing the U.S. automaker's foes to catch up.  GM's current plans for commercialization involve a more scaled back version of the technology called "Super Cruise" which is only semi-autonomous.  
Cadillac Super Cruise
The technology will allow the user to hand off control to the vehicle during long highway drives, but otherwise will drive like a standard vehicle.  GM wrote that the technology will "use a fusion of radar, ultrasonic sensors, cameras and GPS map data, seamlessly integrated" to perform "semi-automated driving including hands-off lane following, braking and speed control under certain driving conditions."

But GM added in on a cautionary note:

The system is designed to ease the driver’s workload on freeways only, in bumper-to-bumper traffic and on long road trips; however, the driver’s attention is still required... because the system will have operational limitations based on external factors such as traffic, weather and visibility of lane markings. When reliable data is not available, such as when there are no lane markings, the system will prompt the driver to resume steering.

In other words, this smart car was only so smart.
 
III. GM Not Alone in Struggles Toward Full Self-Driving
 
The limitations echoed Ford -- currently the world's second largest automaker -- which in 2010 introduced a semi-autonomous parallel parking.  Called Active Park Assist (APA), the system provided so-called "electronic power-assisted steering" (EPAS).
 
In other words, it basically steered for you and tried to tell you how long to press the gas and when to brake via various beeps.  The clear issue, as I saw it during a 2011 test drive was that the technology would do nothing to stop the user from bumping a car in front of or behind it.   And by detaching the driver from part of the maneuver (steering), but not all of it (gas/braking), the maneuver almost felt more dangerous/risky to perform at times.
 
In my experience the system was workable, but at times seemed more frustrating than simply performing the maneuver on your own.
 
EPAS Second 1  EPAS Second Park 2
  EPAS Second Park 3   EPAS Second Park 4
    EPAS Second Park 5  

Toyota had a near identical system -- the parallel park system (PPS), which handled the gas and steering, but not the brake.  Toyota first introduced the system in its Lexus branded luxury models.  It also integrated smart braking technology to avoid objects in driveways, etc. via integrating millimeter wave sensors and the braking system into a collision avoidance algorithm.
 
Ford has been active since, working on an improved version of its lane keeping technology called Traffic Jam Assist, which mirrors GM's "Super Cruise" in providing semi-autonomous highway driving.  That technology should arrive around the same time as GM's.  In the meantime Ford has added perpendicular parking to its portfolio, courtesy of French supplier Valeo S.A. (EPA:FR).  Volkswagen had been using Valeo's Park4U system parking system since the 2011 model year [source].  
 
Last year Ford added one more trick, upgrading the system to be fully autonomous.  With the new system the owner could activate the parking maneuver via a key fob and the Fully Assisted Parking Aid (FAPA) would activate driving the vehicle into or out of a spots.  The feature has proven to be especially useful in preventing scratching to other drivers' doors.
 
IV. Google + Tesla?  Be Afraid.
 
What every automaker should be most fearful of is Google and Tesla teaming up.  The pair has already expressed interest in such a union; as mentioned, Larry Page is a Tesla Roadster owner, as is Google's R&D chief Sergey Brin.
 
Other than Google who is close to full automation?  GM, Ford, Volkswagen, and Toyota are all pursuing the safer, but less exciting route of partial automation for highway traffic.  It is unclear if any of these companies' plans involve merging onto or off of highways; it appears that the automation may solely be limited to highway driving at certain speeds.
 
Nissan Motor Comp., Ltd. (TYO:7201) is taking a similar route, but it might be a bit ahead given that it not only is testing autonomous cars on a Japanese highway, but is merging on and off.  Its autonomous Leaf EV drove itself on the highway trip at speeds of 40-80 km/hour (~25-50 mph).  Nissan has set a target of 2020 for "autonomous" vehicles with Autonomous Drive; however, it is unclear whether that target includes urban/city driving.

Nissan Autonomous Drive
Nissan's Autonomous Drive LEAF EV aims for a 2020 launch. [Image Source: Nissan]

Sweden's Volvo AB (STO:VOLV-A)(STO:VOLV-B) is building fully autonomous driving systems which should merge onto and off of highways by 2017.  During my time with Bayerische Motoren Werke (BMW) AG (ETR:BMW) at CES 2014 I talked to some of their senior engineers and learned that they too were testing automated vehicles.  Those tests are currently confined to the track, but it's possible BMW may field a self-driving coupe in the same timeframe as Volvo.
 
V. Google has Mastered What Its Rivals are Still Trying to Achieve
 
But to be honest, GM is right.  Automakers should be worried about Google.  Because everyone -- and truly every one of them -- is far behind Google.  
 
Google's project began in 2010.  At the time, many believed the power of a supercomputer would be needed for city driving.  Many dismissed the project as a joke or gimmick
 
Two years later Google had already logged 300,000 hours miles of automated driving without an accident.  (Google's fleet consists primarily of retrofitted Lexus RX 450h and the Toyota Prius hybrids.)  Two years later, Google's fleet has logged 700,000 miles of autonomous highway travel.  It's clear that Google's highway driving technology is basically done, versus even Nissan who is still in the testing phase.

Google Lexus


Google has already moved on to finishing a far more ambitious set of algorithms -- routines for city driving.  So far the fleet has logged "thousands" of miles in city driving, leading to vital algorithm improvements.
 
VI. City Driving is Final Test for Google
 
CEO Larry Page recently said that Google has no competitors.  And it truly believes that.  In two years it did more than the rest of the automaker did in the last decade, when it comes to automated driving.
 
The evidence is strong that Google's new self-driving car is not intended as a full fledged design, but as a design focused on urban environment.  Evidence of that is found in its top speed -- 40 km/h (~25 mph) -- which may be sufficient for a crowded city like New York City, but not for highway driving.

Google self-driving car
[Image Source: Google]

Google is so confident in its new design that it's made the bold move of gutting the vehicle removing nearly all controls, including the brake and gas pedals and the steering wheel, something none of the traditional automakers would dare to do.  The new Google X design has two buttons -- Start and Stop.  Program the route and car does the rest.

It's slightly disconcerting and terrifying to think that the Google car's two occupants have no means of controlling the vehicle.  But when you think of it, such feelings are, statistically speaking, overconfidence.  As Google points out 1.2 million people die worldwide in traffic accidents and 90 percent of those are due to human error.

From everything we've seen thus far, you're probably safer allowing Google's algorithms to steer for you.

GM, to its credit has a city-street pod car of its own it's testing -- the EN-V.  By the sound of it that vehicle is still being prototyped, though, and has yet to hit city streets.  GM is developing the curious compact at one of its research facilities in China.

EN-V
The GM EN-V concept vehicle [Image Source: Autoblog Green]

So when all is said and done, it looks highly unlikely anyone will beat Google to the goal of fully automated driving from point to point.  Tesla Motors has probably the best plan -- if you can't beat them, join 'em.

VII. The Android Edge

Even if some other car company -- say GM -- manages to pull abreast of Google in the automation race, it's highly unlikely that its vehicles will ever be as effective at driving.  The reason for that is that Google is continuously mining data -- including location data -- from tens of millions of Android smartphone users across the U.S.  It's almost a given that Google is looking to use this dataset to provide predictive capabilities to its driving algorithm.   

iPhone
Smartphone tracking could give Google beacons to improve its self-driving performance.
[Image Source: Pete Warden and Alasdair Allan]

In other words, Google's cars will be more fuel efficient and safer because they effectively have a beacon on a good deal of American drivers -- regardless of how new or old their car is.  No other company can claim that.


Comments     Threshold


This article is over a month old, voting and posting comments is disabled

RE: no need for total self-drive
By amanojaku on 5/30/2014 12:34:59 PM , Rating: -1
quote:
The point is, if the jackass was in a self-driving vehicle, you'd wouldn't have had something to avoid to begin with.
So your solution would be no manual drivers. Which is not a solution.


RE: no need for total self-drive
By Monkey's Uncle on 5/30/2014 12:52:59 PM , Rating: 2
Why not?


RE: no need for total self-drive
By amanojaku on 5/30/2014 1:09:03 PM , Rating: 2
Because there are only two ways this could happen:

1) A government mandate
2) Drivers voluntarily giving up control

The first will never happen. I can't say how the rest of the world will deal with this, but driving is as much a part of American culture as football and apple pie. The second will never happen, because, again, driving is a part of American culture. There will always be manual drivers, and they will screw up the automated drivers. At least, the current ones; with continued development these things will get better. However, as an experience automated driving does not provide the thrill of manual driving with your own hands and feet.


RE: no need for total self-drive
By SublimeSimplicity on 5/30/2014 1:12:29 PM , Rating: 4
It will be economic pressure via auto insurance.

After a year, when nearly every single accident involving a self-driving car is shown to be the manual drivers fault (and the self-driving car, will have all the info logged to prove it). Insurance on self driving cars will go to almost nothing, while manual cars will go up and up.


RE: no need for total self-drive
By Reclaimer77 on 5/30/14, Rating: -1
RE: no need for total self-drive
By SublimeSimplicity on 5/30/2014 3:31:20 PM , Rating: 2
Smokers want to smoke. Most drivers don't want to drive.

Taking autonomous taxis will cost way less per mile than owning even a junker. It will be what people want for less money.

As saturation gets over 50% and accidents decrease, each fatal accident will get media attention. People driving manually will be vilified for what we consider an honest accident today, because it will be obvious that they chose that to endanger others, even at a financial cost.

IF the tech works, that's what is coming.


RE: no need for total self-drive
By Reclaimer77 on 5/30/14, Rating: 0
RE: no need for total self-drive
By Monkey's Uncle on 5/30/2014 6:00:09 PM , Rating: 3
quote:
Please provide me the survey where all 200+ million registered drivers in the United Stated say they "don't want to drive".


Please provide us the survey where all 200+ million registered drivers in the United Stated say they "do want to drive".


RE: no need for total self-drive
By Spuke on 5/30/2014 6:16:54 PM , Rating: 2
quote:
Please provide us the survey where all 200+ million registered drivers in the United Stated say they "do want to drive".
I guess that means you have no proof to back up what you said. LOL!


RE: no need for total self-drive
By SlyNine on 5/31/2014 12:52:01 PM , Rating: 2
You're the one making the claims... you're the one that needs to provide the proof.


RE: no need for total self-drive
By Monkey's Uncle on 5/30/2014 3:43:56 PM , Rating: 3
Oh man, that is so funny.

What makes you think insurance companies won't? Do you have any idea how much someone driving a dump truck pays in insurance? What makes you think an insurance company won't charge you that same kind of premium once a safer, less dangerous mode of transport comes available to people cheaply?

Self driving cars will have:

1. Less 'at fault' accidents.
2. Less prone to theft.
3. Less often used in crimes as a getaway vehicle (they are pretty freaking slow)

Police departments will love them.

Insurance companies stop being the bad guy for beggaring folks that need to use their cars to go to work.

People who live in cities and abhor public transit (just like someone else I know does) will love them.

People going out and getting shitfaced in a pub will love them because they don't have to rely on 'designated drivers' staying sober, taxis or public transit to get them home. They can take their own car!

The designated driver types in the above will love them because they don't have to BE designated drivers & can get shitfaced right along with you.

Kids will love them because they don't need a license to 'drive' one.

Yes, it is another tool for the evils socialists to punish you :D

(And yes, there are an awful lot less smokers in the world simply from the pressures of society and economics making their smoking habit far less enjoyable)

Don't get me wrong. I love driving my car.


RE: no need for total self-drive
By SlyNine on 5/31/2014 12:35:40 PM , Rating: 2
Where is this mythical city where millions of robots are driving around everyday making perfectly sound judgments and no mistakes are made....

And until one does this is just a hypothesis that exists in your mind. Perhaps some day it will come true but I doubt were as close as you think we are.

So far all these self driving cars have been tested in extremely controlled conditions. It reacted well to getting cut off, please if that's the best test they came up with (the one test a computer SHOULD perform well at) I'm not convinced.

Mixing this technology to aid in a humans driving I'm all for.


RE: no need for total self-drive
By Rukkian on 5/30/2014 2:42:53 PM , Rating: 2
While I don't forsee a time when there are no manual drivers, that does not mean that there will be a ton. Will some people prefer to drive themselves - sure. Just like some people cant stand an automatic vehicle compared to a stick shift. In the end, I think the majority of people use their cars to get from point a to point b (work, groceries, school) and do not really feel the need to be in control.

As for the tech not being able to handle the manual drivers, that is absolutely false. The current vehicles have logged 700k miles with no accidents caused by the automated car, and significantly lower accidents in general, since the computer is always paying attention, and can anticipate what other drivers are going to do, leaving plenty of space, and knowing every escape route possible. There is no way for a human to actually know that much data simultaneously until we grow at least 8 eyes (2 in every direction).

Yeah I know every person is a super human and capable of knowing ahead of time what every other driver is doing if you ask them, but if that were true, we would not have the number of distracted driving, and accidents where a human is at fault in some way or another.

Even if just the 30% of people that are horrible drivers, plus those with less than stellar reflexes (my grandparents come to mind) switch to automated cars, the roads would be much safer, and much more efficient.


RE: no need for total self-drive
By Spuke on 5/30/2014 3:21:05 PM , Rating: 2
quote:
The current vehicles have logged 700k miles with no accidents caused by the automated car
Were all those miles done on public roads? What were the weather conditions? Dry, rain, snow? What were the traffic conditions? Heavy, light, city, freeway?


RE: no need for total self-drive
By Rukkian on 5/30/2014 3:56:24 PM , Rating: 2
I don't have the stats in front of me, but at this point they are allowed (and have been used) on public roads in 3 states that I know of. At this point it has not been tested in snow (I am sure that is coming), but I am sure some rain has been tested.

I don't say it is ready for mass consumerism today, but to say we are no where near being viable (like some here are claiming) seems wrong as well.

I understand that some people are against this and will fight tooth and nail anytime a self-driving car is mentioned, but to a good portion of the public, I don't think the day can come fast enough.


RE: no need for total self-drive
By amanojaku on 5/30/2014 5:05:58 PM , Rating: 3
They're allowed to drive in four states: California, Florida, Michigan, and Nevada. Testing occurred in two states: California (Mountain View and San Francisco), and Nevada (?). Locations that aren't known for heavy traffic or large numbers of traffic accidents.

And I don't think anyone said driver-less cars are not viable as a technology. Even Reclaimer, the most staunch advocate of independence, admits driver-less cars have potential. What people HAVE been saying is today's driver-less technology, even Google's, is unable to reproduce all of the skills the average driver needs. You guys don't want to accept the fact that Google's tests are VERY limited.

Can't blame Google for not testing in the rest of the country (it's illegal), but come on, at least run the thing in LA, Huston, or Austin. Skip Detroit, you'll never get the car back.


RE: no need for total self-drive
By SlyNine on 5/31/2014 1:07:20 PM , Rating: 2
There are plenty of humans out there that will follow that criteria of 700k miles with no accidents/tickets (It's not just one Google car that achieved this.) in fact if mixed results are allowed choosing the right people will lead to figures that BLOW that out of the water.

Plus, you said that a computer can anticipate what other drivers are going to do, mind explaining that one?

Knowing every escape route possible? There are only so many maneuvers available to a car. I contest that a person is more able to predict the outcome of an escape attempt better than current and near future computer software. Which is a prerequisite for figuring out which escape route is better. A human brain is still vastly superior to a computers (of course a lot of people fail to use theirs). At worst your statement is false at best it's only true in part.

Because people drive in the real world... LOTS of them. Until your self driving car reaches the numbers that we see from humans all you're saying is just conjecture and a hypothesis.


"Nowadays, security guys break the Mac every single day. Every single day, they come out with a total exploit, your machine can be taken over totally. I dare anybody to do that once a month on the Windows machine." -- Bill Gates














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki