Autonomous swarming robots, programmed to search like honeybees, could be the best strategy to explore caves on Mars that may harbor life.

Methane traces in the Martian atmosphere point to undiscovered activity — whether geological or biological — lurking beneath the surface. ”Something interesting is going on down there,” said Áron Kisdi, an engineer at the University of Southampton, U.K. “We just need to find it.”

In a paper March 3 in Acta Astronautica Kisdi presents a strategy that he believes offers the best way for robots to search large expanses of Mars for new caves, maximizing search area and minimizing search time.

In 2007 NASA’s Mars Odyssey spacecraft found seven large holes in the surface — too large and deep to be explored with rovers. Odyssey’s resolution isn’t fine enough, however, to reveal smaller caves. And these, if found, could be more accessible to rovers, with shallow entrances and narrower shafts.

“We have sent robots to mars a few times now,” Kisdi said. “But we’ve only seen a small fraction of the planet.”

For his swarm-search strategy, Kisdi envisions using a rolling, jumping robot, Jollbot. A Mars lander would release 40 to 60 swarmbots to autonomously and randomly scout for caves, in the same way bees hunt for nesting sites.

When a robot finds a cave — sensed by a difference in temperature — it returns by the shortest route, the beeline, back to the lander. It wirelessly uploads the cave’s coordinates and temperature readings to the lander. Then it checks the information uploaded into the lander by the rest of the hive and decides to either start a new search or visit a cave discovered by another robot. If it also approves of a spot, it informs the lander and the process begins again.

Within a few iterations, the group comes to a consensus: Either enough bots deem it a good site to point out to mission control, or interest peters out.

The simplicity of this strategy, Kisdi said, allows for the swarming robots to use the bulk of their power on locomotion, rather than programming, and to be cheaper to build. It also ensures that a larger, more sophisticated rover sent in after the bots won’t waste its time on uninteresting places.

“And if you lose a robot,” he said, “the search isn’t over.”

In a report released March 7, a panel convened by the National Research Council asked NASA to give the Mars Astrobiology Explorer-Cacher the highest priority of its large missions. This project, the first of three, would collect samples from the Martian surface for analysis. But the report stresses the mission should only proceed if costs can be cut to $2.5 billion — $1 billion less than current independent estimates. This could give Kisdi’s search algorithm a better chance of, one day, seeing the methane of Mars.

“The hard thing about exploring Mars is providing power to robots on the surface,” said David Beaty, Mars Program Science Manager at NASA’s Jet Propulsion Laboratory. “The surface area of Mars is equal to the surface area of Earth’s continents. Traveling between one cave and another might be a distance of hundreds of miles.”

Beaty isn’t aware of any swarming robot search models under current development at NASA. But, he says, if the robots in Kisdi’s search model were long-lasting and mobile enough, the idea has potential. In Kisdi’s current simulation, 50 swarm robots cover an area of 300 square meters in about five days. The area can be expanded by adding more robots with a longer search distance.

“I’d like to start developing the hardware next,” Kisdi said. “I plan to keep working on the idea until the next call for proposals on Mars missions.”

Image: A candidate landing site for Mars Science Laboratory, Curiosity, captured by the HiRISE telescope. Credit: NASA/JPL/University of Arizona.

See Also:

Citation: “Future robotic exploration using honeybee search strategy: Example search for caves on Mars.” By Áron Kisdi and Adrian R.L. Tatnall. Acta Astronautica, Mar. 3, 2011.