A computer system designed to look for life on Mars has been tested at a site on Earth resembling a Martian region being explored by one of Nasa's rovers.


The trial was carried out in northern Guadalajara, Spain

The system, to be worn by an astronaut, underwent a trial at red sandstone beds in Spain with similarities to Meridiani Planum, where Opportunity rover landed.

Software picks out interesting features and highlights them in real-time in a visor on one eye or a tablet display.

The "intelligent" system can replace geologists' duties, say its creators.

The cyborg astrobiologist consists of a 667MHz wearable computer, a tablet display with stylus or visor, a colour video camera and tripod.

It would provide "augmented reality" allowing astronauts on future Mars missions to narrow down their search for targets relevant to life processes on the Red Planet.

Uncommon features

Patrick McGuire and Jens Ormö of the Centre for Astrobiology in Madrid and Enrique Diaz-Martinez of Spain's Geological and Mineral Institute subjected the system to a four-hour field trial in Riba de Santiuste, northern Guadalajara, Spain.

Part of the reason the scientists chose this site was because in some respects it resembled Meridiani.

Images from Mars orbiters appear to show signs of bleached rings on Meridiani Planum that have been seen at iron-rich geological sites on Earth. This is caused by a chemical reduction process that removes iron impurities from one place and later concentrates it at another site.


Intelligent software picks out interesting features in the landscape

"We don't know for a fact that biology is involved in creating the bleached zones in Riba de Santiuste. But we know that in other red beds around the world, the bleached zones are caused by biological processes," Dr McGuire told the BBC News website.

The wearable computer system picked out similar bleached zones at Riba de Santiuste.

Red 'popcorn'

It even spotted small, dark red concretions or nodules in the rocky beds, similar to the "popcorn" found by Opportunity at Meridiani Planum. These were at the limits of the perception capabilities of the even human geologists.

Images obtained with the video camera are processed by the computer to generate a map of "uncommon" features based on three properties of the image: hue, saturation and intensity.

Uncommon maps for each of the three properties are then combined to create an "interest map" of features in the landscape worthy of further investigation.

Back in the lab, the team compared the points in the landscape that the computer found interesting with those that caught the eye of a human geologist.

The computer agreed with the human 68% of the time. The rate of false positives, when the computer says there is something interesting but the human does not agree, was 32%. The rate of false negatives, when the opposite occurred, was also 32%.

Dr McGuire said the system's performance can be further improved, but added that this test confirms its potential value to planetary exploration.

"You need to have some sort of artificial intelligence expert filter to filter out the false positives. The false negatives mean that you need better computer vision techniques to get you those additional positives," he explained.

Story from BBC NEWS:
http://news.bbc.co.uk/go/pr/fr/-/2/hi/scie...ure/4084058.stm
Published: 2005/06/15 08:11:21 GMT

© BBC MMV