October 14, 2004
Guanaco Camp, Atacama Desert, Chile

Agenda
• Conduct autonomous navigation and science experiments
• Collect science and navigation image sets
• Refine navigation recovery behavior
• Operate with Li-Ion batteries.

Status and Progress
Conducted autonomous navigation experiments. This field season we have conducted over 200 autonomous traverse experiments but today we ran only 15. This was because Zoe made three single-command autonomous traverses over 1km, actually two at about 1300m and one at 3300m, in its most productive day this season at 8940m.

Refined recovery behavior. The most common cause for autonomous traverse to conclude is for the Navigator to find no path ahead. At about 4Hz Zoe considers 15 possible steering arcs, sharp left through to sharp right. Because Zoe only looks about 3-4m ahead it sometimes runs up to an “wall” of obstacles and can find no arc that avoids them. One solution we are pursuing is to use the wide baseline SPI cameras to look for this situations farther ahead but even this is not foolproof. We implemented navigation recovery behavior to get Zoe unstuck and spent some time refining it today. When all path ahead are blocked Zoe backs up along the arc it drove in for several meters and then chooses a different arc. If the path ahead is still blocked it backs up further and tries again. The health monitor and rover executive make sure it doesn’t keep going backwards forever. Today we refined how it chooses new arc and reworked how Zoe switches to that arc to make it turn a bit more sharply. This is working well, and in all of the long traverses today we had several successful navigation recovery actions.

Collected far-field image sets. Throughout the day we logged periodic images triplets from the SPI camera when we encountered interesting terrain. Much of the day Zoe was crossing dry washes so we logged images for future analysis of how to detect and respond to these features while they are still far away and there is time and perspective to choose a best path. Images below show the scene (left), narrow-baseline depth image (middle), and wide baseline depth image (right).

Collected navigation image sets. We also occasionally recorded sequences at various sampling rates from the navigation cameras. These data sets are for future simulation testing of the navigator and for possible work in visual odometry.

Operated with lithium-ion batteries. For most of the field season Zoe has been operating with its reserve lead-acid batteries. This has meant that once it stops receiving sufficient solar energy input, it only has about an hour before it must shut down. We installed the lithium-ion batteries with new safety circuitry, reworked after tests two weeks ago. The batteries ran under minimal load (computing only) and we measured voltages and currents until they were fully charged by the solar panels and the maximum power point trackers cut off. We resumed navigating and logging power from the batteries and throughout the robot. The sun set and Zoe stopped after driving another 4100m autonomously, the cameras could no longer see the terrain ahead. The voltage was about 80.2V, of a maximum of 83V) but charging had probably stopped an hour or so earlier. We spent 30 minutes after dark running full computing load while we down loaded data. Final voltage at shutdown was 78.8V. Our design specification was for Zoe to run four hours with computing and instrument load, but no driving, with zero solar input, and it appears likely that we have met that specification, although further measurements are needed.

Upcoming
• Continued autonomousnavigation
• Tests of the rover executive and fault recovery mechanisms

Weather
Nice but a bit breezy.

Quote of the Day
"Yes, you’ll find that your processes run a bit faster too. And they never get depressed."