American Scientist

Many investigators are faced with the problem of having to conduct field work that is either dull, dirty or dangerous. One solution is rather obvious: assign a graduate student. That time-tested formula has some clear advantages, but now there is another approach worth thinking about—sending a robot into the field. Over the past few years, robotic vehicles have been tested as scientific research assistants on land, under the sea and in the air, and investigators in the earth and planetary sciences are beginning to appreciate what these engineering marvels can and cannot do for them.

Perhaps the most impressive class of mobile robotic minions being developed today is the so-called autonomous underwater vehicles, or AUVs. These small pilotless submarines can be equipped with sensors of various kinds and programmed to carry out observations within the ocean. In some instances they provide the only reasonable means to obtain the desired measurements. One example is the current quest to identify deep-sea hydrothermal vents within the Arctic Ocean. The shifting cover of sea ice there prevents an ice-breaking research vessel from making the necessary systematic surveys using towed equipment. And even if the position of a deep-sea vent were somehow identified, it would be impossible to study it in the usual manner, with a piloted deep-sea vehicle, because of the danger.

A group of investigators at Woods Hole Oceanographic Institution are, however, hoping to locate and photograph hydrothermal vents beneath the Arctic Ocean in the near future using a combination of two AUVs. The first would prospect for hydrothermal sites on the seafloor by crisscrossing the ocean above them and mapping plumes of telltale chemicals in the seawater given off by the vents below. Once the source was pinpointed, a second, more mobile AUV would, according to Robert Sohn, a geophysicist on the research team, do "stuff you can't do off a torpedo," namely hover over the vent and obtain images of the geology and fauna.

For scientific work deep under the polar ice pack, such AUVs may be the only option. But they have proved worthwhile, too, in situations where more traditional oceanographic tools have long been used, such as for mapping the topography of the seabed. For that task, the usual approach is to employ acoustic transducers mounted directly to the hull of a research vessel. When a more detailed view of the seafloor is desired, acoustic equipment can be towed underwater. But it proves quite difficult to maneuver equipment towed from a ship at the end of a lengthy cable. One cannot, for example, make the sensor package turn on a dime or approach the bottom too closely without risking collision. Even when all goes well, surveying in this fashion proceeds slowly, because a long cable cannot be forced through the water at normal cruising speed.

AUVs overcome all of these limitations and so can produce highly detailed charts of the seafloor. Woods Hole's Autonomous Benthic Explorer ("ABE") has, for example, mapped volcanic features on the seabed around the East Pacific Rise. This locale has been studied for many years; still, "ABE blew everyone's mind with these bathymetric maps," says Sohn.

Surveying the Earth in one way or another also constitutes an important application for many of the robotic vehicles now winging through the air. Unmanned aerial vehicles, or UAVs as they are called, vary enormously in size: One of those flying for science sports a wingspan of only 3 meters, whereas another stretches for more than 75. Judith Curry, an investigator at the University of Colorado, Boulder, and her colleagues have used the smaller variety extensively over the past two years to obtain meteorological observations over the Arctic Ocean and to gather detailed views of the sea ice below. Curry also flew this type of robotic aircraft out of Florida last fall to study the hurricanes brewing nearby. Her plan was to fly the UAV into a hurricane at extremely low altitude, where no sane pilot would be willing to venture. Her scientific objective was, however, thwarted after the attacks of September 11th, when the Federal Aviation Administration forbade aircraft to operate without radar transponders. The diminutive vehicles she uses are too small to carry such equipment. "This is the big deal for UAVs—regulatory issues," says Curry.

Others engaged in work with UAVs echo Curry's sentiments. Richard Blakeslee, an atmospheric physicist at NASA's Marshall Space Flight Center, is engaged in a demonstration program using the General Atomics Altus UAV. (This aircraft is a high-altitude variant of the Predator, a remotely operated surveillance plane being used by the U.S. military.) His test flights are limited to the airspace around El Mirage, California, where General Atomics has a flight facility and has worked out special arrangements with the FAA for operating its pilotless plane over land.

Like Curry, Blakeslee intends ultimately to study intense storms off the Florida coast. But his motivation for using a pilotless vehicle rather than a conventional research aircraft is different. Flying at great height over a tempest is not particularly dangerous. It is, however, dull—circling a storm for hours on end, as would be required to follow it from inception to final collapse.

The Altus can remain airborne for 30 hours, and it is better suited to this mission than most high-altitude research aircraft for a second reason: It flies quite slowly and can thus loiter over a thunderhead. The high-altitude jets of the sort that have been outfitted for research zoom by a storm in a couple of minutes and then must slowly turn around to make another quick pass. As Blakeslee notes, "Most of the time, the plane is somewhere else." Blakeslee thus sees the slowness of Altus as a great advantage. And eventually, it may be cheaper to operate than piloted aircraft, although that is not the case yet. "Right now, these UAVs are not cost effective, if you want to know the truth," explains Blakeslee.

Saving money is also not a reason right now to use a robotic vehicle on land. Consider the wheeled Nomad robot, which investigators from Carnegie Mellon University brought to Antarctica two years ago to locate and identify meteorites. "They had gotten from NASA an obscene amount of money," says Ralph Harvey, a planetary geologist at Case Western Reserve University who directs the National Science Foundation's Antarctic Search for Meteorites program. Applying Nomad in the hunt for Antarctica meteorites probably represents the most advanced scientific application so far for robotic vehicles on land. And Nomad, with its elaborate artificial intelligence, displayed an impressive ability to find and classify meteorites.

But as Harvey explains, that project was merely a demonstration of the technology, not a realistic strategy for aiding scientists in the field. "It takes time away from humans to nurse-maid a robot right now," says Harvey. He points out that while Nomad indeed found a half-dozen or so meteorites during its Antarctic trials, one person on the support team "found 60 meteorites in his spare time." So for Harvey at least, the traditional solution still looks better than a robot for finding meteorites: "A graduate student can do that 100 times faster—and maybe cook me dinner too."