--by Graham Ryder

Photo by Laszlo Keszthelyi

Marsokhod in the field. The image shows a gully or crack and its opposite wall, as well as the slope that the Marsokhod is on.

TEST DAY 1. NASA AMES RESEARCH CENTER, CALIFORNIA-- Through the stereo cameras of the Marsokhod rover, we can see an exposure of the far wall of a gully. We think we are looking at dark basalt flows towards the bottom, with paler ash layers on the top. But can we be sure? Can we see what we need with these images? We ask for some high quality monochrome stereo television images and an even higher quality color image with a different camera. Because the color camera is fixed, unlike the stereo cameras, which can pan to see in any direction, it is difficult to point the Marsokhod here so that the camera has the appropriate tilt; the slope we are on is too steep. So we have to back off that request.

The Missions and the Teams

For three days in February 1995 teams of scientists tested an unmanned robotic rover in making geological observations. The Marsokhod is a six-wheel drive, Russian-built vehicle. McDonnell Douglas and NASA Ames Research Center are mainly responsible for its development and operation as it now exists.

Marsokhod was at Kilaeau in Hawaii making traverses assisted by volcanologist Laszlo Keszthelyi (University of Hawaii) and his team. The team at Ames included engineers and controllers, led by Butler Hines (NASA Ames), and the geology group I belonged to.

We were trying to find out how reasonable our remote observations and interpretations of the geological features are, as well as figure out how a team can organize itself, make decisions, and produce strategies for obtaining information. At bottom it was a geologically-based training run for a potential mission to send a rover to the Moon and try to understand its volcanic features. (Immediately prior to our study, another group of geologists had gone through a similar exercise for a rover mission to Mars, which included the time-delay that would be involved in communications over the Earth-to-Mars distance.)

Kilaeau has varied volcanic characteristics. Its topography is rugged, with rifts and gullies. The ground surfaces include ashy materials and other wind-blown fragmental debris. On the Moon there would be more regolith and less bedrock. But for this test, we weren't trying to precisely simulate lunar terrain; we were assessing our ability to observe and successfully interpret volcanic features in general--and whatever else we might see--through the eyes of a rover.

Field area of Southwest Rift of Kilaeau with the prominent lava flows and the little rifts (cracks and gullies). The main caldera (Halemaumau) is in the background.

Our geology team started out as Cass Coombs (College of Charleston), Larry Crumpler (Brown University), and me, all with experience in field geology, volcanology, and planetary studies. Jeff Taylor (University of Hawaii) had planned the traverses. We used aerial photos to simulate orbital and descent images. One of us was in charge of decision-making while the others provided advice and geological input from images. We rotated the functions daily.

Three people were not enough to keep pace, so we enlisted the help of Jayne Aubele (Brown University) and Carol Stoker (NASA Ames). On a real mission we would need far more. But we were here to learn, not simulate. We made decisions about where and when we would use instruments for chemical and mineralogical analyses on a mission, but this test had no such instruments.

TEST DAY 3. NASA AMES RESEARCH CENTER, CALIFORNIA-- By the start of the third day, we have become a field commander, a chief analyst communicating with the commander, and two analysts. We spend some time finding where we are (the field team has told us roughly where we "landed"). Nobody mentions the surface materials, as we are too busy panning for identifiable landmarks. Then we recognize lava with "dirt" that is probably a wind-drifted fragmental material (later we are told that the "dirt" is indeed windblown tephra--fragmental deposits).

We start to traverse even though we don't know exactly where we are. At first we take no detailed images for scientific analyses, but use one taken for navigation. We approach an outcrop and examine it. Shiny surfaces look like glassy surfaces on lava. We can tell it is solid and hard. We argue (or discuss?) alternatives. Is it glass reflecting sunlight? If we look from another direction what does it look like? A human field geologist would move rapidly and answer some of these simple questions very quickly; with the rover each movement is a carefully considered option. We decide that a (time- consuming) chemical analysis would be required here. We interpret holes as vesicles in the lava. And so we make our inferences, collect images, imagine that we would have made the chemical analysis, and discuss where we are off to next.

Photo by Graham Ryder

NASA Ames control center, showing Cass Coombs observing the TV screens, coinciding with photo above.

Ruminations & Results

The idea of unmanned vehicles roving over planetary surfaces, operated from Earth, is hardly a new one. However, over recent years the concept has received tremendous stimulus. Funding for planetary exploration using humans will not be easily forthcoming (it is now more than twenty years since any living creature went further than low earth-orbit), whatever the balance of virtues and capabilities of robots and humans is. The tremendous advance in robotics and telepresence that the computer and miniaturization has enabled makes the capabilities of robots and their instruments potentially much greater than they were even a decade ago.

These advances have prompted some to state that all our major objectives in planetary exploration can be met using robots. But we have not yet really tried out such exploration. Attention has focused on the mechanics and operation of a robot rover and not on the quality of the observations and inferences. Little attention has been paid to the real-time cognitive aspects of knowledge production (i.e., humans thinking) in geological field work. Unlike a static vehicle such as Viking, a rover continuously and rapidly requires many decisions about what to look at, with what instruments, and when.

At Kilaeua, we did quite well on many observations, according to the debriefing when it was all over. We were able to identify flow fronts, vesicles in basalts, tension cracks in the ground, and stratigraphic relationships among basalts, and not just because of prior knowledge. We were able to define different types of surfaces. The pace of movement was sometimes frustrating (a human can scan and select very rapidly) and the optical limits sometimes equally so. The test emphasized the basic reality that learning has as much to do with forgetting as remembering: eliminating the irrelevant from the synthesis.

The team had varied opinions on how close this all came to actually being in the field; I never lost the feeling that I was encumbered by the limited vision and was never able to feel that I was actually there, but others came close. We still have a lot to do to define team structure, team strategies, and decision- making, but this test gave us a lot of confidence that we can use such a rover to make excellent observations on other planets.

It is inevitable that planetary exploration will include robotic roving, such as Mars Pathfinder. We need to know how to do it well. For our next tests we would want some improvements: better real-time imaging, continuous knowledge of the heading and tilt of the rover, and faster image-processing and informed feedback. We would make more panoramic sweeps as we drive.

We have to accustom ourselves to the difference between rover vision and our own vision: the much lower viewpoint and much wider instantaneous field of view of the rover tricked us into thinking objects only a few meters away were a hundred meters away. And we have to learn how to question our observations, make sure we complete our observations, and learn how to communicate rapidly with each other. But everyone needs more than one class when they take Drivers-Ed! We are excited at the prospects of another next lesson!

(Dr. Ryder is a Staff Scientist at LPI.)