The VIPER Atacama Mission - 10 July 97

Carlos Ernesto Guestrin, from the desert

This morning I looked into some calibration problems and found that the turn-table that rotates the sensor head was loose. I opened the turn-table, breaking the manufacturer's seal (oops!), and found that the internal mechanism was loose.

The good news is that even with large errors in yesterdays panoramas, the multiple panorama algorithm was able to estimate a position within 450 meters of the GPS measure.

The plan today was to go North and drive back taking panoramas along the way. The day was beautiful, with a bright blue sky contrasting the colours of the desert. Here you can find almost any colour, except green! Although, I was told there are some plants in hidden locations, the only non human living things I have seen were a butterfly (!!!) and a locust. I must say, with all the dirt, I am not sure we really qualify as humans.

At first sight, the desert seems dull, however, it is very colourful. There are different shades of brown in the mountains to the West, ranging from almost black to almost white. The "Salar de Atacama" is white from the salt, with intense red mountains around the edges. On the far East, the Andes stand high covered with snow, with perfectly conical volcanos.

I drove North and took my first panorama around 23 degrees 9 minutes South. When I was about to take the second panorama, the power failed. At first, I thought the battery was discharged. Latter I found that the power inverter had died. I brought two and the other was at the camp site, so I had to return earlier than I expected.

I arrived just in time for lunch, the food is great!!! A mining camp nearby is providing food for the group. So far, it has been excellent homemade Chilean style. We always have a soup, main dish and dessert. I am eating more than what I used to eat in Pittsburgh!

After lunch, I went out to take more panoramas. I managed to take two just before sunset, which is about 5:30pm. I was amazed by beautiful red sky. They say "red sky at night, shepard's delight", well I don't know how delighted a shepard can get out here, but it sure won't rain tomorrow. This is the driest desert in the world. In average, it rains once every hundred years.

Once I got back to the camp site, I started processing today's data. The processing follows three main steps:

  1. generating the panorama, using the image registering algorithm;
  2. detecting the skyline, using a image processing-based algorithm;
  3. matching the skyline from the panoramas to those rendered from a digital elevation map of the region.

The best match is then compared to the position estimated by the GPS and the uncertainty is determined.

On processing the two panoramas I got just before sunset, the position was determined with uncertainties of 326.7 and 564.4 meters in a search area of 1600 square kilometers. Furthermore, when the multiple panorama algorithm was used, the position was nailed down to 193.8 meters.

These results are excellent, specially because the the resolution of the digital elevation map we have of the Atacama is of 90 meters and many mountains are not on the map.

  back to Viper system

This work has been conducted at the Robotics Institute at the School of Computer Science, Carnegie Mellon University. It has been partially funded by NASA; Fabio Cozman has a scholarship from CNPq (Brazil). We thank these four organizations for all their support.