Education & Workforce Development and Environment and Government and Technology

Accepting a Grand Challenge: Jones' robotic startup aims at military market

May 2, 2005

Within the next 10 years, the U.S. Department of Defense hopes to fully automate a third of its ground vehicles. Indianapolis-based high-tech entrepreneur Scott Jones has plans to one day sell the robot pilots the military needs to accomplish that mission.

But before he can build a business capable of attracting serious venture capital, he has to build a robot that can drive a Jeep Rubicon across 175 miles of the Mojave Desert in less than 10 hours. And he has to do it by October.

"We've already focused on the right problems," Jones said. "I guarantee you if we perform well in this, we'll have all sorts of opportunities. And come October, we're going to have that seed-level startup."

The deadline was set by the federal Defense Advanced Research Projects Agency, or DARPA, which has issued a "Grand Challenge" to robotics experts around the nation. After it narrows the field, the top 20 teams will place their vehicles at the starting line on Oct. 8, enter the final destination coordinates, and watch them head off toward the horizon. Teams can't interact with their robots until the race is over. The first one to cross the finish line earns $2 million.

This fall actually marks the second DARPA Grand Challenge. And if last year's results are any indication, the military's ultimate ground-fleet automation goal remains distant. Last year's top team, Carnegie Mellon, barely made it past the seven-mile mark. But that was better than the rest of the pack. A few hundred feet was all most entrants could manage.

"Grand Challenge is probably the right word," said Dan Hirleman, head of Purdue University's School of Mechanical Engineering. "Air vehicles have a lot less problems than ground vehicles. The variety of terrains and obstacles and unknowns that you have to deal with, we're just kind of in the beginning of [addressing] that problem."

Jones, who worked as a research scientist in MIT's Artificial Intelligence Laboratory before making his fortune in voicemail patents, admits robot vehicles are in their infancy. It's much like the earliest days of the aviation industry.

"Look at a picture of a Wright [Brothers] flyer next to a 747," he said. "There's the challenge."

But the potential market is enormous. Already, he points out, the military is handing out contracts in the field worth tens of millions of dollars. And in the future, the armed forces won't be the only ones with a demand for robot-driven vehicles. Jones envisions farms driven entirely by automation, or cars that can safely take elderly or handicapped people to the grocery store.

The key, Jones said, is not to start from scratch. Rather than attempt to duplicate technology that already exists, Jones and the other 78 members of the volunteer Indy Robotics LLC team are focusing on the problem of how the robot "sees" its environment in order to make steering decisions.

"We're not trying to reinvent the wheel. We're going right at the really hard problems," Jones said. "Rather than do what the majority of teams are doing and focusing on things that have been done before, we're standing on the shoulders of all of that."

That's why the vehicle platform is a Jeep Rubicon. To actually turn the wheel and work the brakes, Indy Robotics incorporated technology from locally based Ahnafield Corp. Founded in 1968, Ahnafield converts vehicles so highly disabled people can "drive by wire." In this case, the "disabled" driver of the Jeep Rubicon is a computer.

Founder Bruce Ahnafield considers his involvement with the project a groundlevel investment in a brand-new industry. And he's confident Indy Robotics will outperform last year's DARPA field.

"With the people who are involved and the technologies involved, I've got a very high level of confidence that we'll succeed," he said. "I wouldn't have invested the time and money in the project if I didn't believe in it."

For now, Indy Robotics is simply attempting to make it to the Mojave. There are 118 teams vying for the competition's 20 slots, said Doug Traster, head of Indy Robotics' technical team.

The future of Indy Robotics as a business isn't necessarily contingent on winning the race. Simply performing well should be enough to attract some attention, Traster said.

"Even if we don't succeed in winning the championship itself, the head of DARPA will be there personally," he said. "If someone can actually succeed at this task, it should be possible to get millions in government contracts. That's what we're hoping to bring to Indiana, not just $2 million but tens of millions to build [robots] here."

In addition to Ahnafield, the Indy Robotics team includes participants from Indiana University, Purdue and Rose-Hulman Institute of Technology. Hirleman, for example, has 120 electrical and mechanical engineering students working on their own vehicle for the 2006 race.

"Normally, in our educational projects, we don't have that obvious tech transfer link," Hirleman said. "Here's one where at the same time our students are having this capstone design experience on campus, there's some useful things that can be spun off. And it may be that some of our students may be interested in joining the [Indy Robotics] team."

Thus far, Indy Robotics has spent less than $200,000 in cash, Jones said. The volunteer structure of the infant company allowed it to avoid the challenges earlystage startups find most challenging.

A full-time business model, Jones said, would have cost $5 million to $10 million just to attempt, and it would have held a great deal more inherent risk.

This way, Jones said, Indy Robotics can establish proof of concept before it approaches investors. When it does, it will have a solid story to tell.

"I didn't need to raise money, and I didn't need people to quit their jobs. All these people came together to make this dream come true," he said. "There's nothing quite like this on the East or West coasts. It's our own Midwestern 'do it different' kind of thing."
Source: XMLAr01900.xml
ADVERTISEMENT

Recent Articles by Peter Schnitzler

Comments powered by Disqus