Posted on 09/26/2005 4:59:01 PM PDT by BenLurkin
MOJAVE - It's a long way from the woods and waters around Ithaca, N.Y., to the desert scrub of Mojave, but it's a trek worth taking if you're looking to test an unmanned vehicle capable of crossing rugged desert terrain all by itself. Oh, and it could net you a $2 million prize and untold prestige.
Looking to win the DARPA Grand Challenge race this fall, the team from Cornell University has set up shop at a Mojave Airport hangar this summer, perfecting and testing their vehicle in an environment similar to the one they will face come race day.
Sponsored by the Defense Advanced Research Projects Agency, the Grand Challenge on Oct. 8 is a contest for unmanned robotic vehicles to begin and end in Primm, Nev. The vehicles will be required to travel approximately 150 miles along a route that will be announced only hours before the race begins, using just their onboard navigational aids and sensors to avoid obstacles.
The team whose vehicle successfully completes the route in the shortest time under 10 hours will win the $2 million purse.
Eight Cornell team members arrived in Mojave on Aug. 1 from the campus in Ithaca for the final construction and testing of their entry, and will remain here through the qualifying trials in September and the race itself.
To test their Jeep-like robo-vehicle, the team looked for a location that offered inexpensive living and working arrangements in addition to room to put their vehicle through its paces, and work space to house it all.
"Mojave is basically perfect," said senior Brian Schimpf , with terrain that closely resembles last year's course.
The Mojave connection was made through alumni Dan DeLong, an engineer at Mojave Airport-based XCOR Aerospace. DeLong arranged for hangar space for the project and testing area.
So many contestants attempted to test their vehicles on last year's course between Barstow and Primm, Nev., that DARPA closed off the entire area. The boundary line for the forbidden zone lies just a few miles south of Mojave, Schimpf said.
The basis of the Cornell team's entry is a rugged, off-road military vehicle by ST Kinetics. It is used by the Singapore and other armed forces.
"It's meant for really fast off-road use," Schimpf said. "It can go over anything, handle anything."
Other teams have used standard SUV bodies for their entries, but the Cornell team came to the conclusion that these did not have the stability needed for the bumps and ruts of the rugged course and in negotiating obstacles.
"This, you take (the bumps) at any speed and you hardly feel them," Schimpf said.
Powered by a diesel engine, the vehicle carries a stack of four computers - all off-the-shelf hardware - encased in a metal box mounted on heavy-duty motorcycle shocks to protect the computers from the rigors of the terrain.
This computer "brain" controls the vehicle, constantly evaluating its situation and making necessary adjustments.
The control program continuously runs simulations to determine what will happen with its next move; "basically it sees the future," Schimpf said.
"We call it 'danger checks,' " teammate Aaron Nathan added.
With a solid chassis beneath it, the most difficult aspect of the autonomous vehicle is the sensor system that allows it to detect and avoid obstacles along the route.
For the Grand Challenge, entries must navigate the entire course on their own, without human interference.
"We just hit the 'go' button and it's gone - 150 miles by itself," Nathan said.
Teams will be given the coordinates outlining the route shortly before the race starts. Following that route and staying within the required bounds is the easy part, Schimpf said.
The difficulties arise in the unknown obstacles, natural and manmade, that will litter the race route.
To avoid these pitfalls, accurate sensing is vital to the vehicle's - and the team's - success.
Cornell's entry uses a trio of laser range-finders mounted across the front of the vehicle, above where the hood would be on a normal car.
These sensors - labeled "Larry, Curly and Moe" - project lasers into the path ahead. By measuring the time it takes for the beam to reflect back, the sensors can determine if there is an obstacle in the path and how far away it is.
The sensors offer great accuracy, up to 10 cm, Schimpf said.
The middle sensor, mounted slightly higher than the other two, is set on a swivel so it can guide the vehicle around corners.
On either side of this sensor are two cameras, offering a stereo view of the terrain ahead in much the same way human eyes function. By calculating the differences between the images captured by these cameras, the computers on the vehicle can judge distance.
"That's something we developed ourselves," Nathan said.
The technology of the sensors themselves is only one part of the equation; integrating them with the entire navigation and control system and making the whole thing work in harmony is the challenge, Schimpf said.
Nathan is the team's test driver, sitting behind the wheel during test runs to quickly override the automatic control system if something goes wrong.
Taking it out on the first test drive was "one of the scariest things ever," he said. "You just see the car is moving and you're not doing anything. It's like something out of a science fiction movie."
DARPA initiated the Grand Challenge in order to stimulate interest and development of autonomous vehicles, which the Department of Defense views as a critical element of future military capabilities.
Such vehicles could prove important for tasks such as convoys transporting supplies across the Iraqi desert, which are vulnerable to attack.
Last year, in its inaugural run, the Grand Challenge did not produce a winner. The $1 million in prize money was doubled for this year's challenge.
The Cornell team's project began just more than a year ago, during last year's competition. About 35 to 40 students began working on the framework of their team, enlisting university support and raising funds.
They also studied the work of the 2004 teams, analyzing the methods they used and determining how and why those efforts failed.
"That was the main reconnaissance we did on the other teams," Schimpf said.
The vehicle itself arrived in January, when work began in earnest in constructing the entry.
Some team members who traveled west during the summer returned to school for the start of fall semester classes. Schimpf and Nathan, along with two other teammates, are taking the semester off to concentrate on the project. They will receive credit for their work as an independent research project, however.
Both, entering their senior year at Cornell, have previous experience on similar engineering project teams.
Electrical engineering major Nathan spent two years with a robotic soccer team. The Grand Challenge project gave him a chance to indulge more than one passion: cars and robotics.
"If you were to take everything I find interesting in life, it would be this car," he said.
Schimpf, who is majoring in industrial engineering but is "really a computer science major at heart," worked on a hybrid car project previously.
"This was incredibly cool and no one had done it," he said of his decision to join.
So far, the team has made a successful run of as much as 2 miles, enough to test the individual systems they have been evaluating.
The problem under scrutiny is the vehicle's ability to adequately see obstacles in its way.
"It basically has drunk vision," Schimpf said.
"It's 90% done, which means we still have nothing," he added.
The project would not be possible without numerous donations of equipment, funds and aid. These contributions account for about 80 to 90% of the project in total, Schimpf said. The students estimate the project cost between $350,000 to $400,000.
The first challenge for the team will be meeting 42 other teams in the qualifying matches, from Wednesday to Oct. 6, at the California Motor Speedway in Fontana. From there, 20 qualifying teams go on to the final race.
(and they're banned in several states because of "dirty politics")
...notwithstanding they have tons of power and get 27MPG with B10 biodiesel
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.