Autonomous cars turn the corner

The other day I rode in a car that no one was driving.

Well, OK, there was a guy named Mike in the driver’s seat. But he wasn’t touching the steering wheel or pedals. That was left to a system connected to a 2.7 GHz Intel Quad-core blade server—and essentially three backup systems—stacked in the cargo area of the Ford Explorer in which I rode. Riding along were Randy Iwasaki, chief deputy director of California Department of Transportation (Caltrans) and Chairman of the Intelligent Transportation Society of America, and Federal Highway Administrator Thomas Madison, Jr.

The self-driving Explorer, outfitted by the Southwest Research Institute (SRI), pulled onto 11th Avenue in New York City, drove two blocks south, made a U-turn in front of a parked van, weaved through barricades, stopped at a stoplight, crept past a parked van, looking for pedestrians that might step out in front of it, made another U-turn three blocks later, turned right, and parked itself in front of the Jacob K. Javits Convention Center.

Mostly it worked flawlessly. If you count driving like an overly cautious, newly minted teenage driver as flawless. At least, we didn’t hit anything, never errantly crossed a lane line, and Mike was never forced to take over.

Two radar units were fitted on the front bumper, and the SUV kept track of its position via GPS satellites. The radar could identify objects beside the car and in front, informing the computer to enable safe lane changes. SRI says it has had the machine up to 80 mph under autonomous control, and it expects that is about the limit of its technology. One caveat is that GPS mapping data is not precise enough: programmers have to tell the machine exact coordinates of things like lane dividers and curbs.

Along with the second-place-winning Stanford University entry from the Defense Advanced Research Projects Agency (DARPA) Urban Challenge that was on hand, this Explorer showed it is possible for cars to drive themselves, even in a challenging urban environment. I have no doubt that the day will come when cars can drive themselves in many situations. But the jerky ride up and down 11th Ave. (especially when it came to the brakes) showed there’s a lot more development work needed before computers can substitute for human vision and judgment. For now, the technology is promising, though like a teenager, supervision is recommended.

—Eric Evarts

Comments

[DT]

I have watched a few TV programs on this new technology and I know that it will soon be common place. But I'm curious, what is the ultimate goal for this type of advancement in the automotive culture. Especially in the United States, we as americans have a special type of connection with our automobiles. Driving has always been like an extension of our unalienable rights, the free spirit, the constitution, simply the freedom to go anywhere at anytime with very little needed but the desire to get behind the wheel. As much as we complain about traffic, the truth of the matter is Americans love their cars and love the freedom of driving, proof of this is the failure of most public transportation systems around the country in our major cities. My readers enjoy working and modifying their automobiles and more importantly establishing that one on one connection with their cars. And this is just not possible unless your hands are gripping the steering wheel with your foot on the pedal in anticipation of a burst of torque as you step into that gas. My point is and I think it reflects the attitude of many in this country and that is we like to drive. And if you think that we are willing to give it up so that some computer program can do it for else, well it's a far stretch I believe. It may look cool on TV or in a Box Office Movie but in real life?

[BradYourCarGuy]

I think the preceeding commentor is onto something. Driving is much more than a means of transportation. If all we wanted was to get from here to there we would all be driving the most economical car possible. We drive because we like to drive. This technology, however, does have important usage in things like safety features to keep us from turning our favorite car into a hunk of twisted metal. The same technologies that are being used here can be used for anti-crashing (?) systems, something that might takeover to keep us from plowing into that tree or oncoming car, or to take over when we accidentally fall asleep.

[Brad Templeton]

The implications of this technology actually extend quite far beyond driving. While driving for fun will still exist, I think it will take place largely in rural areas like mountains or windy roads. The need to drive is generally overestimated; New Yorkers largely give it up and are no less American for it. And when you look at the big implications -- that human driving kills 45,000 people per year, with a cost of $230 billion according to the NTSB, and also wastes 50 billion hours of time -- you need to really love your driving to say it's worth that.

But it goes further. Robot cars have the potential to make small, efficient, short-range cars marketable, removing U.S reliance on foreign oil. Is that worth it?

I've been studying this for a while and have built a large web site of essays on the topic which expound more on these ideas. It can be found via the link on my name or pointed to from http://robocars.net

[Eric Evarts]

DT,

As I understand it, there are three ultimate goals here:
1) Improve automotive safety by relying on computers to do things like stop the car in an emergency, and watch out for pedestrians. Studies have shown they are more reliable and suffer fewer attention lapses than human drivers. :)
2) Reduce congestion by increasing the "throughput" of our existing roadway system. Nobody thinks we can just keep endlessly building more travel lanes. Reducing congestion would also reduce fuel consumption and reduce emissions.
3) Satisfy a new generation of drivers who have grown up in teh age of congestion and pollution from cars and have indicated they are less attached to them than, say, baby boomers. These young folks are now telling focus groups, at least at General Motors, that they would rather spend their travel time texting friends, getting work done, or watching a movie than having to do the driving themselves.
Clearly, this technology won't be universal. But it might go a long way toward meeting these goals at least in urban areas. And when people want to go out of their way to go for a drive and explore where their cars can take them, they should still be free to do that on less-congested rural byways.
Hope that helps.

[sarahsmile90]

Call me old school, but I like driving. Or call me a control freak, but I like to control my vehicle. I appreciate the need for safety features, but robot cars take it a bit to far for me.