Special Coverage


DETROIT – Self-driving “autonomous” vehicles could be on public roads as early as 2013, says a research scientist at Carnegie Mellon University's Robotics Institute.

In five to seven years, expect to see autonomous vehicles capable of navigating freeways, says Chris Urmson, member of a joint Carnegie Mellon-General Motors Corp. team that won the $2 million first-place prize in a prestigious 2007 autonomous-vehicle competition sponsored by a research-and-development arm of the U.S. Department of Defense.

Vehicles capable of navigating urban grids are 10 to 20 years away from reality, Urmson says during a panel discussion here at the Convergence Transportation Electronics Conference.

Sensor sophistication, validation capability and assuring performance in adverse environmental conditions are key stumbling blocks, but Urmson says these are not insurmountable.

“Technology is making great steps,” he adds, admitting the notion of driverless vehicles sounds as fantastic as “The Jetsons” cartoon series.

Urmson's remarks are greeted with little skepticism by fellow panelists, though Continental AG executive Peter Reith warns the technology likely will launch in phases. Expect the first autonomous vehicles to be restricted to “special lanes,” says Reith, management board member-chassis division and safety.

Autonomous vehicles represent a new chapter in the auto industry's history, and “we are standing at the beginning,” he tells Ward's.

Adaptive cruise-control technology manages a vehicle's longitudinal movement, Reith says, while lane-departure warning systems portend efficient and safe lateral movement. The next hurdle is enabling the vehicle to handle unexpected situations.

“Are vehicles able to tell the difference between a fencepost and a person?” Urmson asks.

Session moderator Larry Burns, GM vice president-research and development and strategic planning, suggests the engineering challenges are less onerous than the legal ones.

“Technologically, (an autonomous vehicle) is possible,” Burns tells Ward's. “But we have a lot of concerns about privacy and liability.”

The former is an issue because the navigation component of autonomous-vehicle operation requires satellite tracking. And liability presents a potential nightmare if an autonomous vehicle is involved in an accident. “Everybody will say the driver is in control,” Reith says.

Burns also notes the autonomous vehicle debate is a global one. “So we shouldn't think that a solution flows right through the U.S.,” he says.

Paul Brubaker, U.S. Dept. of Transportation administrator-research and innovative technology, reminds there will be regulatory implications, as well, although he declines to predict what federal mandates might be considered.

However, he does say 90% of the vehicles on the road today must be replaced by 2030 if vehicle connectivity is to fulfill its potential.

John Waraniak, Specialty Equipment Market Assn. vice president-vehicle technology, says this could be achieved by 2020 if auto makers were more amenable to open architectures. That would drive more innovation, which would feed demand for personalization, something consumers have shown a willingness to pay for.

“Customerization,” as Waraniak calls it, “is no longer a trend, it's a requirement.”

Anup Sable, vice president-automotive, KPIT Cummins Infosystems Ltd., warns of traffic congestion as autonomous vehicles compete with one another for real-time traffic data, not unlike the way servers are swamped when accessed by too many computers.

Sable elicits the biggest laugh of the session when he reminds that certain segments of society already have solved all the headaches associated with developing autonomous vehicles.

“That solution is called – having a driver,” he says.