CONTACT: Stanford University News Service (650) 723-2558
Stanford students win national aerial robotics competition
STANFORD -- A team of Stanford engineering students took first place in the Fifth Annual Aerial Robotics Competition on July 6, a victory based on the first use in that contest of the Global Positioning System for satellite navigation.
The competition, held in Atlanta at the Georgia Institute of Technology and sponsored by the Association for Unmanned Vehicle Systems (AUVS), requires that flying robots locate six small metal disks scattered in a six-foot ring, capture them one at a time, carry each of the disks over a tennis net and drop them in a second six- foot ring -- all without direct human control.
According to event organizer Rob Michelson, president of AUVS, the event provides the students with important experience in solving real-world engineering problems. He said it also has increased public interest in autonomous aerial vehicles, which can be used to reduce the risks of combat scouting in wartime and can help lower the cost of routine pipeline, forest or highway inspection work.
The three-man team from Stanford's Aerospace Robotics Laboratory included Stephen Morris, a former research associate in aeronautics and astronautics at Stanford; Andrew Conway, who just received his doctorate; and doctoral student Bruce Woodley. Faculty advisers were Stephen M. Rock, associate professor of aeronautics and astronautics, and Robert H. Cannon, the Charles Lee Powell Professor of Aeronautics and Astronautics.
Several days before the contest, team member Morris said that the best previous performance was hovering for 20 seconds, and the team was pretty certain it could do better.
It did. Its autonomous flying machine -- a heavily modified gas-powered model helicopter -- became the first entry not only to hover under control, but also to move to the ring containing the metal disks and pick up one.
"We're delighted that everything worked perfectly," Rock said. "There are a lot of things that can go wrong in a competition like this. We were very fortunate."
A total of 16 teams from five nations qualified to enter the event. Of these, a dozen showed up for the contest and all but a few got their machines flying. A balloon-based entry from the Technische Universität Berlin from Germany took second place by demonstrating stable hover and navigation around the field. The third place entry, from the University of Texas-Arlington, achieved a stable hover.
Going into the competition, the Stanford team members knew that they could not complete all the tasks that were required. They had designed a system to pick up the disks on command, but it was too heavy for the helicopter to lift. So they rigged up a simple capture device: a wooden cross with four permanent magnets at each end dangling from a long string. Such a device can pick up a disk when it comes into contact, but cannot release it.
On its first flight, the Stanford helicopter disqualified itself by picking up two disks, rather than one. On its second effort, it failed to pick up any disks. But on its third try, it successfully picked up a single disk and carried it over the net to the drop-off ring.
Because the Stanford entry did not complete the entire task, the team was awarded a first prize of $7,000, rather than the full $10,000 that it would have received if the robot helicopter had managed to pick up all six disks and drop them in the ring.
Stanford's autonomous flying machine was the first entry to use satellite navigation to control its movement. Its onboard computer relied on radio signals from Global Positioning System satellites passing overhead to update its position, including altitude, five times per second. Although the military GPS system originally was designed to provide civilian users with positions accurate within 10 meters, the students used a technique called differential carrier phase GPS to provide the computer with position information accurate within centimeters.
"Most of the entries have had to use gyroscopes and a number of different sensors to achieve the same result," said Conway, who spearheaded the effort. "But our robot has only one type of sensor, which has no moving parts."
The robot helicopter sported four GPS antennae, two GPS receivers and a 486 computer. This system controlled the helicopter's orientation as well as its motion. The students first determined the precise positions of the boundaries of the competition area, and the location of the starting area, pick-up ring, net and drop-off ring. They then programmed the onboard computer to instruct the helicopter to climb to a set altitude; calculate the direction to the pick-up ring; fly to the pick-up ring; drop to the correct altitude so that the pick-up magnets would just drag on the ground; quarter the pick-up area for so many seconds; climb to an altitude high enough so that the magnets would clear the net; calculate the direction to the drop-off ring; and fly to the drop-off ring.
The students' major concern before the competition was the amount of lift that their helicopter would produce in the hot, humid conditions in Atlanta. In converting the helicopter to fly autonomously, they doubled the weight of the original kit. Although they changed the fuel mix, fine-tuned the engine and added larger blades, they were uncertain how much weight the craft would be able to lift. Another potential problem would have developed if the competition area had been located near tall buildings that could block the satellite position signals. Neither proved to be a problem.
Next year, the competition will be held at Epcot Center in Orlando, Fla. "We have been challenged to come back next year, and if we can get the funding we intend to do so," Rock said.
Photo available electronically by anonymous ftp at 126.96.36.199 under the filename "Aerialrobot.tiff". Caption information is contained in associated file "aerialrobot.readme". Please use your full e-mail address as your password.
This is an archived release.
This release is not available in any other form.
Images mentioned in this release are not available online.
© Stanford University. All Rights Reserved. Stanford, CA 94305. (650) 723-2300.