Suggested Searches

5 min read

Swarmies Shuffle Through Field Tests

Months of lab work has led to this chilly day – by Florida standards – with four small, wheeled robots moving around the parking lot outside the Launch Control Center while their leader, Kurt Leucht, keeps electronic tabs on them using a laptop. He carries the laptop around as he tracks each of the four machines, occasionally tapping one off an obstacle or looking at the vehicle’s line of sight to figure out what its sensors are seeing.

Kurt Leucht works with the swarmies
“Swarmies” project lead Kurt Leucht works with the “Swarmies” during field tests at Kennedy.
NASA/Dmitri Gerondidakis

Together, the robots are known as the “Swarmies” and it’s not their hardware that makes them noteworthy, but rather the coding each carries in its silicon brain that make them search the same way ants do. Each of the robots has its own camera and a set of hazard-avoidance sensors. They are rolling around looking for barcode sheets and leaving digital trails to tell the others when a whole lot of bar codes are found in one place.

“We’re trying to prove that there’s more efficient ways of searching than some other more common ways,” Leucht said. “It works really well for ants and we think it could work just as well for robots.”

Leucht isn’t controlling the robots in the way a radio-control hobbyist does – he’s not doing their thinking for them. Instead, he’s letting the software he and his team have been working on for months do the work of operating each robot independently. Besides, that’s the way mechanical creatures like this would have to operate on Mars if they are to be effective resource gatherers.

Working with computer engineer Caylyn Shelton at Kennedy and a research team at the University of New Mexico Biological Computation Lab, Leucht is using this parking lot test and dozens more just like it to try to see whether a search method based on foraging behaviors is more effective and productive than a conventional approach of scouring every square inch of an area or a purely random search. It is the same approach used by ants for eons to find and collect food and material.

The approach could be put to use in the future in service of astronauts making the journey to Mars.

One idea among many is to dispatch a corps of small robots capable of searching the Red Planet for water-ice and then digging it up for processing into breathing air and rocket fuel. The robots – purpose-built, flightworthy machines loaded with software like the coding Leucht is working on – would arrive at Mars months or years ahead of astronauts and use the lead-time to build up a storehouse of resources that would be waiting for the explorers from earth.

With each robot being small and weighing less than 10 pounds, a large fleet of searcher/gatherer machines could be sent into space on a single launch. With 100 robots tooling around on the surface, it also wouldn’t matter so much if several broke down because there would be plenty more to do the work.

“When we eventually send people to Mars, it would be best if we use in-situ resource utilization techniques, which is a fancy scientific term for living off the land,” Leucht said. “A swarm like this could be really useful in detecting resources in a mission like that which needs to go out and collect resources because it’s an autonomous system.”

All that is far down the road for Leucht right now, though. Before the robots can roll around on Mars scouting about for good areas to mine, he has to make sure these little earth-bound versions do everything in real life that the simulation program said they’d do.  

“Working with real robots is a lot harder than working with simulated robots,” Leucht said. He’s a software engineer by training, so this project is his first encounter with robotics. “There have been a lot of bugs that we have had to overcome, both hardware and software. When we got them outside into the parking lot for the first time, there were a lot of things that they did that we didn’t expect them to do. It took a lot of troubleshooting and debugging and a lot of detective work to figure it out. Now we’re down to the nitty-gritty of just letting them run these outdoor trials and collecting data so we can prove whether this biologically inspired search technique is better than other search techniques.”

The results are promising, Leucht said, and improvements are constant.

This effort has been underway for almost a year and the goals of it are all but complete, Leucht said. The next step is to find out whether a proposal to expand the simulation capabilities of the software will be funded. If so, then the broader community of robotics experts can get involved and perhaps advance the technology faster and fine-tune the search approach into something that can be made into a swarm of robots suited for Martian work.

“It’s kind of human nature to want to improve them and put more money into the robots, but we want to prove that this will work with simple robots and cheap sensors,” Leucht said. “We want to make the software more accessible for testing by other universities and organizations since not too many are using biologically inspired behavior and genetic algorithms.”

Such an approach may also open the doors to an increased role for the software in other robots built for work on this planet such as search-and-rescue and land mine detection and elimination.

“There’s all kinds of earthbound uses for a project like this,” Leucht said. “It’s really exciting to think of the possibilities.”

This project is supported by NASA’s Center Innovation Fund, part of the agency’s Space Technology Mission Directorate, which is responsible for innovating, developing, testing, and flying hardware for use in future missions.