Smart Mars Rovers

A scientist talks about a new tool for space exploration — a robot that thinks like a geologist. When a space probe sends data from Mars, the radio signal can take up to 20 minutes to travel back to Earth.

And if mission controllers send a command back to the probe, it might take another 20 minutes to reach Mars. This delay makes it tough to explore Mars by remote controlled rover.

Martha Gilmore: It would be nice for the rover to be able to pick out things that are important scientifically on its own.

That’s Martha Gilmore, a geologist at Connecticut’s Wesleyan University. She and her colleagues are developing software that will allow Mars rovers to identify interesting minerals and take samples on their own. Such minerals include carbonates — rocks that usually form in the presence of liquid water and might hold evidence of primitive life on Mars. Robots using these programs can already pick out carbonates from a variety of minerals.

Martha Gilmore: But this is not Mars — these are data collected on Earth. So the next step is to deal with martian conditions — such as the spectrum of the atmosphere, coatings on rocks that we’d expect — and these are the types of things that we’ll be simulating in my lab over the next few years.

Martha Gilmore and her colleagues at NASA’s Jet Propulsion Laboratory are working to program a rover to think like a geologist.

The Sojourner Rover that landed on Mars aboard Mars Pathfinder (http://mars.jpl.nasa.gov/) in 1997, and the rovers was scheduled to fly aboard the 2003 Mars Exploration Rovers (http://marsrovers.jpl.nasa.gov/home/index.html) mission need constant control from scientists and engineers on the ground. This requires a huge team of controllers on hand around the clock for the duration of a mission. They analyze incoming data, send commands, and wait for results during the 8- to 40-minutes it takes a radio signal to reach and return from Mars.

Depending on where they are in their respective orbits, Earth and Mars can be as far apart as 378 million miles — or as close as 78 million miles. The commonly cited 8-minute round-trip communication time is possible only when Mars and Earth are closest together.

This time-consuming and labor intensive process greatly reduces the efficiency of any Mars Rover mission, so scientists at NASA want to teach rovers to think for themselves.

Martha Gilmore said, “It would be nice for the rover to be able to look around and say ‘These are the things that I think are important.’ And when we wake up the next day, those priorities will be there when we get to work and are able to process those and go on from there.”

To accomplish this task, NASA’s artificial intelligence experts are writing neural net algorithms that can identify the types of minerals geologists find most important on Mars. Geologists are keenly interested in any minerals formed in the presence of water, because any possible life on Mars would have almost certainly needed water to evolve and survive. While some believe that life could have once existed all over Mars, any evidence of life would be confined to rather specialized geologies, Gilmore explains. “Lets say a little bug is swimming around. You’re only going to capture that bug if you have minerals that are forming quickly that can entrap and entomb those bugs. And for fossils on Earth, many of them were formed or preserved because you had very rapid sedimentation able to preserve things quickly.”

The kinds of rocks that are candidates for harboring potential signs of life are carbonate minerals, which on Earth generally form in shallow seas, and hydrothermal minerals – rocks formed underground in interactions between magma and water. Gilmore and her colleagues are doing research in places such as Yellowstone National Park – a hotbed of hydrothermal activity – to identify properties common among hydrothermal minerals. Programmers can then write algorithms that will help rovers recognize such rocks, Gilmore says.

The goal is to build rovers that can identify these types of minerals from less-promising ones, set their own geologic priorities and begin sampling important rocks on their own. Well-rested scientists on Earth can then sit back and receive the results during a regular work week.

What you have in your mind?