University Of Florida Robotic Car Will Compete Nationally

June 6, 2005

GAINESVILLE, Fla. — A University of Florida-built robot car was selected today to vie for a spot in a $2 million race across the desert.

“Team CIMAR,” representing more than a dozen faculty, students and outside engineering experts who helped design and build “NaviGATOR,” was selected among 40 semifinalists in the Defense Advanced Research Project Agency, or DARPA, Grand Challenge 2005.

Winnowed from an original field of nearly 200 entries nationwide, the UF team is now headed to qualifying events at the California Speedway in Fontana in late September and early October. Just 20 teams will emerge from qualifying to compete for a $2 million prize in the main race, set for Oct. 8 at an as-yet-undisclosed location in the Southwest.

“We are thrilled to have this opportunity, and we plan to go all the way,” said Dave Armstrong, project manager for Team CIMAR, the UF Center for Intelligent Machines and Robotics group that built the car.

The team has been working on NaviGATOR since early last fall. The effort comes on the heels of the first Grand Challenge in March of last year. No one came close to winning that event, which required fully robotic, or autonomous, cars to complete a 142-mile course from near Los Angeles to near Las Vegas across the Mojave Desert. The leading team, from Carnegie Mellon University, ran just seven miles of the course. The UF team finished eighth after traveling roughly one mile.

With the benefit of that experience and a completely new car, this year will be different, Armstrong said.

Constructed on a box-like blue tubular steel frame, NaviGATOR looks something like a tricked-out Land Rover. But although the car has many factory-made parts, including a Honda Civic engine, it is completely unlike any manufactured car.

The big difference: computers and sensors that give NaviGATOR the ability to sense and respond to the terrain around it without any human intervention.

The DARPA event will require cars to run a narrow, preplanned course replete with obstacles and potential competition-ending hazards such as ditches, cliffs and boulders. To meet the challenge, NaviGATOR has 10 PC-based computers on board, all housed in an air-conditioned box that rests on air springs designed to isolate it from vibrations.

The computers gather and interpret information from numerous on-board navigation and hazard-avoidance technologies, including Global Position and Inertial Navigation systems, cameras, and ultrasonic and infrared sensors. The information then goes to actuators, or automatic controls, that steer, accelerate and brake the car. A seat and traditional pedals on the car allow a person to drive NaviGATOR. But when the computers take over, the pedals move up and down by themselves like player piano keys.

In a recent demonstration, the car drove itself briskly around a test track near Gainesville, successfully avoiding hay bales and barrels placed in its path. When it approached an obstacle, it slowed as if considering the possibilities, then turned its wheels and proceeded around it.

In its longest run ever, the car on May 20 completed more than 24 laps around a dirt track, successfully avoiding 120 barrels and 24 hay bale obstacles. When done, the NaviGATOR had traveled 12 miles at an average speed of 7.5 mph.

If more than one car completes the competition, the prize will go to the fastest. That points to the main challenge: running the course as quickly as possible.

“It’s all about speed,” said Carl Crane, a professor of mechanical engineering and director of the CIMAR lab. “Can you trust your sensors, or are you going to get false readings?”

Mike Griffis is the chief executive officer of Eigenpoint, a robotic systems engineering company near Gainesville and one of the three main sponsors of NaviGATOR. He said he and his staff have spent hundreds of hours working on the project.

“It’s a tremendous challenge. It’s ‘The Big Challenge’, and I like challenges,” he said.

Team CIMAR has invested about $250,000 in cash and equipment in NaviGATOR. If the costs of engineering labor were added, the car would likely be a close to a $1 million project, Armstrong said. The car’s other main sponsor is Autonomous Solutions, a Utah-based company that builds robots for civilian and military purposes.