Desert RATS Test Robotic Rover
by Henry Bortman
Moffett Field CA (SPX) Sep 30, 2005
Until earlier this year, when President Bush announced an ambitious blueprint for space exploration, NASA had no plans to send humans back to the moon, or to Mars.
But that didn't stop an intrepid group of scientists, based at NASA's Johnson Space Center in Houston, Texas, from investigating technologies that would be needed for such a mission.
Every year for nearly a decade, they have trekked out to remote desert locations to conduct research on equipment and procedures that might some day be used by off-world explorers. The skunkworks project, known as Desert RATS (Research and Technology Studies), has just completed its eighth field season, on a barren cattle ranch near the rim of Meteor Crater, some 40 miles outside Flagstaff, Arizona.
It's hard to find places on Earth that simulate Martian conditions. But the Arizona high desert comes close enough for the experiments the Desert RATS team conducted during the first two weeks of September.
According to Joe Kosmo of Johnson Space Center, who has led the Desert RATS effort since its inception, Meteor Crater is an ideal test site because if you "strip away the vegetation, put the atmospheric pressure at 100,000 feet, and put the sun a little farther away, essentially you're encountering the kind of terrain you'd see on Mars," Rough, slightly hilly desert hard pack, with an assortment of rocks and boulders strewn about. And dust (red, of course). Dust everywhere, blowing around in heavy gusts, making dust devils and coating everything in site.
This year's two-week field test focused on the interaction between a pair of "astronauts" (actually, space-suited scientists) and a rover named SCOUT (Science Crew and Operations Utility Testbed).
When astronauts travel to the moon or Mars, they will be going to do pretty much the same things a geologist does when exploring a field site on Earth: walking around, observing land formations, taking pictures and collecting rock and soil samples. But unlike on Earth, where even at the most remote locations, help is usually not far away, off-world explorers will have severely limited resources. Moreover, they will want to investigate as much terrain as possible, so conserving energy to focus on scientific tasks will be important.
"We're trying to augment the human-machine cooperative working relationship so that the machines can do a lot of the tedious tasks," says Frank Delgado, project lead for SCOUT. For example, by using SCOUT to drive explorers "to the location where they need to do their science, when they get there, they're a lot fresher."
SCOUT looks like an oversized dune buggy. Its design is loosely based on the Moon Buggy used by astronauts on the Apollo 15, 16 and 17 missions back in the 1970s. Its seats are built to accommodate two passengers wearing bulky spacesuits. Its joystick and computer touch screen are optimized for easy use by heavily gloved hands. And it is tricked out with an eclectic array of cameras, speakers and a host of communications gear. Its maximum ground speed: 6 miles per hour.
SCOUT had its first real-world test during the 2004 Desert RATS field season, but in that initial shakedown it was driven under manual control. "We basically focused on having somebody drive it from onboard," says Delgado. "So if you needed to get to a crater or somewhere to collect some rocks, they would jump onboard, and they'd drive it."
This year the RATS team tried out several automated modes of operation, including tele-operation, voice commands and gesture recognition. The field tests, which were highly successful, were the first ever that involved such complex interaction between humans and a semi-autonomous robotic assistant.
One type of tele-operation involved operating the rover in real- or near-real-time from a remote location. The operators used a joystick and a set of switches, knobs and buttons to control the rover as though they were onboard. Delgado's team was able to tele-operate SCOUT in this way both from a Desert RATS command center located about a half-mile from the test site and from a control center about 1200 miles away at the Johnson Space Center in Houston.
This is a feasible approach for operating a robotic rover on the moon from a lunar base or even from Earth. It could also be used by humans operating a rover from a local command base on Mars. It wouldn't be possible to tele-operate a Mars rover from Earth in real-time, however, because it takes too long for radio-command signals to get from one planet to the other.
Earth-based scientists could tele-operate a rover on Mars using batch commands, however. For example, "You can say, Go to waypoint 1, take a picture; go to waypoint 2, take a panorama; go to waypoint 3; and then come back home.' And it will do that automatically," says Delgado.
This is similar to the way in which mission controllers operate the Spirit and Opportunity rovers currently exploring Mars. But sending commands to Spirit and Opportunity requires first going through a laborious process of translating science-team requests into a sequence of arcane commands in a specialized language that the rovers can understand. SCOUT can understand direct voice commands.
Perhaps the most intriguing mode of operation that Delgado's team successfully tested was a procedure known as "human following," in which the rover followed one of the scientists as he explored on foot. To initiate human following, the scientist stood in front of the rover and said, "SCOUT, follow me." Onboard the rover, Delgado explains, is a pair of stereo cameras and a computerized shape-recognition system that knows "what a person should look like.
It locks onto them and as they walk around in front of the vehicle, the cameras will swivel to that direction. If they walk away from the vehicle, the vehicle actually follows them, like a pack mule. So if they're doing some sort of geological expedition and they're going to go out a half mile or a mile, the vehicle will be right there for them to get back on, instead of them having to walk all the way back."
Delgado's engineers also got SCOUT to respond to commands issued in the form of gestures. One of the scientists stood in front of the rover and put out his arm, as though signaling for a left turn. SCOUT recognized the gesture and turned on its lights. When the scientist bent his arm at a 90-degree angle, SCOUT turned the lights back off.
This was only a proof of concept, but it holds great promise. For example, says Delgado, an astronaut could point to an interesting rock and say, " Take a picture of that.' [The rover] will see where their finger's pointing and it'll turn the cameras in that direction and take a picture."
Although future missions to the moon and Mars will utilize vehicles with SCOUT-like capabilities, SCOUT was built as a testbed, and not as a prototype of a rover for a future mission. "There's a lot of basic functionality in the vehicle," Delgado explains, "that's there to support some of the concepts that we're trying to develop - the autonomous operations, tele-operations, obstacle avoidance, human following."
But features like rubber tires with air in them, or the rover's aluminum frame "would never be flown to the moon or to Mars." Moreover, the computer systems on SCOUT are built from off-the-shelf components. Hardware used on vehicles intended for spaceflight have to be specially designed for protection against dust, intense radiation and other severe conditions.
Still, the lessons learned in testing SCOUT will no doubt be applied to designing whatever rovers do get built to accompany humans on their return to the moon, and eventually to Mars.
http://www.spacedaily.com/news/mars-robot-05b.html