Smarter Spacecraft: Science-Hunting Software for Robotic Explorers
By Tariq Malik
NASA’s robot space explorers are getting smarter, due in no small part to the human-written software that powers their mechanical minds. On Mars, for example, the twin exploration rovers Spirit and Opportunity received a boost in their autonomous driving skills during recent software update by mission controllers. But a team of NASA and Arizona researchers are pushing spacecraft self control even farther, giving their robot the power to conduct its own scientific investigations using software that automatically scans for interesting targets.The satellite, Earth Observing-1 (EO-1), used NASA-developed software to make its own studies of an Antarctic volcano and is capable to tracking flooding events around the world before ground controllers tell it to, researchers said. Since the volcano - Mount Erebus - is known to be active, it was a prime test target for the smart software, they added.
"Not only does this software detect activity, it starts making its own measurements before researchers even look at the data," explained Ashley Davies, lead scientist for the experiment at NASA's Jet Propulsion Laboratory (JPL) in Pasadena, California. "It does what a human on the ground would do."
The study is part of NASA's Autonomous Sciencecraft Experiment (ASE), a program to enhance a spacecraft's ability to conduct science investigations without specific direction from human handlers. In addition to JPL's volcano-targeted software, the project also includes flood and ice tracking software developed by university researchers from the University of Arizona and Arizona State University, Davies told SPACE.comThe approach can cut lengthy time lags inherent to taking an observation, transmitting it to the ground for study, and later decisions by scientists to direct the satellite to take further measurements.
"It takes time to download that data and have scientists pore over it," said James Dohm, a research scientist with the University of Arizona who participated in the study, during a telephone interview. "It can take days to weeks and even longer."
A smarter satellite
The ASE software aboard EO-1 consists of three parts that work together to spot science targets autonomously.First, a set of science algorithms contains specific instructions on what to look for in incoming data. On EO-1, that includes directions to look for volcanic heat signatures and flood activity in images as they are collected by the spacecraft's Hyperion imaging spectrometer. But the process could be tailored to fit any mission. Scientist searching for a specific mineral on the moon, for example, need only algorithm that would search through a lunar satellite's data until it finds that mineral's spectral signature, Davies said.The ASE software also uses an onboard planner computer program to prioritize which science investigations EO-1 should conduct on its own and take control of the satellite, as well as aa spacecraft command language (SCL) to communicate with onboard instruments. A potentially large benefit of the ASE software is its ability to circumvent communication time lags associated with the huge distances between planets. Signals sent by the Galileo spacecraft during its time at Jupiter, which is about 580 million miles (934 million kilometers) from Earth, took more than 50 minutes to reach ground controllers. The even farther Cassini probe to Saturn took about 84 minutes to send messages across the 934 million miles (1.6 billion kilometers) between Saturn and Earth when it arrived at the ringed planet last week.
Instead of waiting to process data, and in the meantime flying past your object of interest, ASE software aboard a spacecraft could keep tabs on the target until researchers have a chance to see the data, Davies said.
Knowing what to look for
Planetary researchers agreed that smarter science software could be useful for future Mars rover or Jupiter probe missions, but the approach is very dependent on a previous knowledge of the area under study."It really depends on the specific mission and whether you know enough to give those smarts," said Joy Crisp, project scientist for the Mars Exploration Rovers mission at JPL. "It certainly is a new tool for scientists."Crisp told SPACE.com that in order to be useful the ASE software would have to look for known science features, such as hematite in a follow-up mission to the Opportunity's Mars landing site Meridiani Planum. In a Mars mission designed to seek out organic material - which would use the abundance of carbon as a marker -, a drilling lander could use the software to switch on science instruments if it finds high concentrations of the element, she added.
Davies' volcano-targeted ASE software is a good match for Earth observing spacecraft because researchers already have a wealth of knowledge about their home planet. Preexisting data on the volcanic activity and even the knowledge that Mount Erebus was an active volcano factored heavily into the software's successful test, researchers said.
With the distances to other planets so vast, it's most likely that a Jupiter-bound spacecraft, for example, would not immediately use ASE software during its scientific mission."Once you're at Jupiter, everything is pretty much scripted," said Ed Gamble, software engineer for NASA's proposed Jupiter Icy Moons Orbiter mission currently under development. "You would have already determined how you want to do your science by then."But since ASE software exists in lines of code rather than moving parts, it could easily be uploaded to the spacecraft if a specific need to conduct automated science was required. The software doesn't even have to be used during a mission to sift spacecraft data for science nuggets. Its algorithms could mine spacecraft data sets after a mission is over to seek out additional interesting observations.
"This is really a golden age of space exploration and discovery," Dohm said, adding that he has spent years participating in Mars mapping studies. "But I can't go there with my field boots, so the next step is to send even smarter robots instead."