nai

The two NASA rovers, Spirit and Opportunity, which are speeding their way towards a January rendezvous with Mars, are arguably the most advanced robotic spacecraft ever sent to explore another heavenly body. But to the robotics researchers working on the Life in the Atacama project, Spirit and Opportunity are already history.

This group of hardware and software wizards, based at Carnegie Mellon University’s Robotics Institute are designing rovers that, they hope, will roam the Martian surface, not in this decade, but in the next. Their goal: autonomous robots with sufficient onboard intelligence to explore for days at a time without human intervention.

The development of these new robotic capabilities is being field-tested in Chile’s Atacama desert, which stretches east from the Pacific coast high into the Andes mountains. At its heart lies perhaps the most arid region on Earth – an area so dry that even extreme-adapted microbial life has difficulty surviving there.

The researchers, whose work is funded jointly by NASA’s ASTEP (Astrobiology Science and Technology for Exploration of Planets) and Mars Technology Program, hope to accomplish two related goals during their three-year effort. Most immediately, they hope to chart life’s ability – or inability – to survive in the Atacama. This will help biologists understand the limits of life on Earth. But in addition, they hope to develop rover hardware and software, as well as life-detection technology, that can be applied to future missions to Mars.

From the Arctic to the Equator

Their first field season was completed earlier this year. The rover used in the field experiments is called Hyperion. Originally deployed in the high Arctic, Hyperion was designed, literally, to follow the Sun. Because the Sun is in the Arctic sky all day long during the northern summer, Hyperion was designed to be able to operate 24 hours a day, by maneuvering so that its fixed solar panel constantly pointed sunward.

The current research is being conducted near the equator, but the same Sun-tracking software used in the Arctic comes in handy in the Atacama as well, to maximize the distance that the rover can travel during daylight hours.

The basic function of this software, known as TEMPEST (Temporal Mission Planner for Exploration in Shadowed Terrain), is to develop long-range navigation plans for Hyperion. TEMPEST was developed by Paul Tompkins, a doctoral candidate in robotics at CMU. The software takes several factors into account, all of which contribute to developing an energy profile for the rover.

Sunlight, as mentioned above, is a critical factor.

“If you have to drive entirely through shadow, for example, to get from point A to point B, and there’s not enough energy onboard the battery, it will disallow that as a viable path, and it will have to go around the shadow to achieve the objective,” says Tompkins.

Another factor is large-scale terrain.

“Say, for example there are two different routes from point A to point B and one takes you over a large hill and the other one goes around it. You can actually trade off the extra distance you might travel to go around the hill versus the extra energy you might have to impart to the rover to get it up and over the hill,” Tompkins says.

Stored in the rover’s memory is satellite imagery of the terrain it is traversing. This helps scientists select a long-range target for the rover.

“We’re actually using a priori information derived from satellite data,” says Tompkins. “So we have a map, what we call a digital-elevation model, that we’d gather ahead of time.” On Mars, similar information is available from the data collected by MOLA (Mars Orbiter Laser Altimeter), one of the instruments onboard the Mars Global Surveyor.

Also stored on Hyperion is software called SPICE, designed by the Jet Propulsion Laboratory, which calculates “the positions and rotational orientations of the major bodies in the Solar System over time. So you can predict the azimuth [direction along the horizon] and elevation of the Sun at any point on a given planet,” Tompkins says. Combining the elevation and solar-position data makes it possible to predict what areas of the local terrain will be in shadow.

One Step at a Time

Applying this information to the target location, TEMPEST develops a set of shorter-range goals, known as way points, which are separated by 30 to 100 meters. TEMPEST feeds the way points, one at a time, to another software program, the local navigator, whose job it is to make sure the rover avoids small obstacles, such as rocks, along the way. The local navigator is similar to the software built-in to the Spirit and Opportunity rovers.

As it reaches each way point, Hyperion stops and re-examines the surrounding terrain.

Tompkins explains that “it’s difficult for the [science] team to understand what ‘s happening if you take a picture at point A and then a picture at the end at point B, one kilometer away. It could be a very different perspective physically. So if you take intermediate photographs, and then in addition to that maybe take some other scientific information along the way, geologically or biologically, then you can get a feel for how those quantities change over the course of the traverse.”

Although only about 20 such sets of measurements were acquired during the first field season, in the second and third seasons (in 2004 and 2005), many more such measurements will be made. This will enable scientists to begin mapping the gradient of life in the Atacama, both in coastal regions that receive moisture from nighttime salt fogs and in the more-arid interior.

In the field season recently completed, the Life in the Atacama team achieved a significant milestone: an autonomous one-kilometer (0.6-mile) traverse. (For comparison, Spirit and Opportunity are expected to travel a total distance of about one kilometer each.)

“We had been shooting next year to try and be able to have the robot navigate one kilometer on a single command,” says David Wettergreen, a research scientist at CMU and the technology lead for the Life in the Atacama project. “And because we were getting very good results, going 6-, 7-, 800 meters in a command cycle, we actually were able to test at least one traverse of a kilometer on a single command to the robot. I don’t know that that type of distance has been accomplished by a planetary rover in testing before, and it puts us in a good position for next year, when we are going to do that again and again and again.”

This time around, the rover was not, strictly speaking, fully autonomous. Its control software was running on an off-board processor, which sent its commands via a wireless link to the rover. And many of the scientific instruments were not yet integrated onto the rover. Instead, a team of scientists tagged along behind the rover and performed experiments and measurements manually.

“But next year,” says Wettergreen, “we’ll have those instruments integrated with the robot. So as it does these traverses, it will be able to periodically collect those data sets, or at least it will be able to do collection on a much more regular basis.”

What’s Next

Plans are already underway for the second field season, which will take place in the fall – that’s northern-hemisphere fall – of 2004. The Life in the Atacama team has ambitions plans for its second expedition.

“The goal in the second field season is to visit two sites,” says Wettergreen. “One that is, again, in the coastal range, and then one that is in the heart of the desert, the hyperarid region. And to traverse longer distances. So we set a minimal goal of [a total of] about 50 kilometers (about 30 miles) of traverse. “The big change, I would say, though, for the second field season, is that we anticipate to have the various science instruments integrated with the rover.”

Tompkins also has some pet projects in mind. One of these is to write software that will enable the rover, before it shuts down for the night, to “select a spot that will be best to hibernate in. So my hope is that I can build in the capability to select maybe a hillside that receives early Sun. So it might distinguish between areas that receive sunlight only later in the day from areas that might receive early sunlight. And so it might get a jump start on operations on the following day.”

He also wants to build in allowances for uncertainty.

“If you have to get to a certain place by a certain time, but there’s an uncertainty in how long it takes, then you might shoot to start earlier,” he says.

“Or, in the case of falling into that fault,” Tompkins adds, referring to a problem that occurred in this year’s field season when Hyperion wandered dangerously close to a steep drop-off, “if you knew that there was some uncertainty where the fault actually was, then you might choose not to get so close to that feature, to protect against potential position errors.”

Meanwhile, in a separate project not directly related to Life in the Atacama, Wettergreen hopes to put Hyperion to use in its original polar configuration.

“We have for five or six years now been working on a lunar-rover initiative that would look for ice and search the polar regions of the Moon,” says Wettergreen. So that concept of the vertical solar powered rover seems very appropriate to that environment.”

This configuration conceivably could be used one day to explore the polar regions of Mars, which some scientists believe is a prime target in the search for signs of extraterrestrial life.