CES: GM on evolving smarter cars
This article was originally on a blog post platform and may be missing photos, graphics or links. See About archive blog posts.
A car that can drive itself -- now that’s a gadget Americans would love. But the self-guided Chevy Tahoe that General Motors was showing off in the parking lot across from the Las Vegas Convention Center wasn’t exactly a production model.
Developed by a team that also included Carnegie Mellon faculty and students, the modifications that enabled the SUV to win last year’s DARPA Urban Challenge in Victorville, Calif., aren’t going to be available as factory-installed options any time soon. And really, does anybody honestly want a LIDAR rig mounted on their roof?
Still, Bakhtiar B. Litkouhi, a manager in GM’s research and development division, said some of the technologies that enabled the Tahoe to pilot its own way through 60 miles of simulated city traffic are starting to creep into vehicles today, and will evolve into more powerful versions in the next few years.
‘We can easily envision a freeway-type autonomous driving,’ Litkouhi said. The building blocks are already there: adaptive cruise control that automatically slows and accelerates a vehicle to keep pace with highway traffic, and lane-monitoring technology that uses tiny cameras tied into a vehicle’s power steering to keep a car within the lane lines. GM is expanding its adaptive cruise control to work all the way down to a complete stop (currently, it cuts out around 20 mph), which makes it more suitable for Los Angeles freeways (ahem).
Several manufacturers offer cars that can maneuver themselves into a parking space; Litkouhi envisions vehicles that can drive themselves to nearby parking lots, then return when summoned. GM is already working on pedestrian detectors -- a safety feature that European regulators are pressing manufacturers to add within the next few years -- and something Litkouhi called a ‘virtual bumper’ that stops a car automatically when the driver isn’t braking quickly enough to avoid striking something nearby.
Another piece is vehicle-to-vehicle communication. Putting a bunch of road sensors into a car can be expensive, but some of the cost can be avoided by having each vehicle tell the ones behind it when it’s slowing down or encountering a slippery road. ‘If the vehicle ahead is saying it’s braking, I don’t need a radar,’ Litkouhi said.
A quick ride in the passenger seat of the autonomous Chevy Tahoe (dubbed ‘Boss’) showed that there are some things technology can’t yet replace. For starters, Boss didn’t exactly have a feathery touch on the gas or brake pedals. Chris Urmson, technology director for the Carnegie Mellon team that worked on the DARPA Challenge with GM, explained that the Tahoe’s driving style was tuned for racing, not comfort. More seriously, when the SUV came to an intersection, it had no way to wave on other drivers or otherwise communicate its intentions, aside from its turn signals.
‘Driving is a social activity,’ and drivers’ gestures play an important role in that, Urmson conceded. ‘Cars can’t do that yet.’
-- Jon Healey
