Sheepdog Trials Provide Inspiration for a New Autonomous Robotic Swarm Algorithm

Researchers from the Georgia Institute of Technology have turned to a classic example of controlling swarm behavior to come up with a new approach to handling large numbers of autonomous vehicles, drones, and robots: herding with a sheepdog.

"Birds, bugs, fish, sheep, and many other organisms move in groups because it benefits individuals, including protection from predators," explains associate professor Saad Bhamla of the team's work. "The puzzle is that the 'group' is not a single organism. It is built from many individuals, each making local, imperfect decisions."

In the case of sheepdogs, their ability to herd a flock of sheep comes from what Bhamla explains is the "selfish herd behavior," where individuals near the edge of the flock will instinctively move to the center for protection when threatened by a predator. "Shepherds exploit that instinct using trained dogs," the researcher says.

Examining hours of footage from professional sheepdog trials, the researchers discovered something seemingly counterintuitive: larger groups are easier to control than smaller ones, with greater numbers of herd members feeling protected in the center. Smaller groups, by contrast, have members flicking between "follow the group" and "flee the dog" behavior. "That switching behavior makes the group unpredictable," co-lead Tuhin Chakrabortty explains.

The team's breakthrough comes in applying these lessons to autonomous robotics. Rather than taking the typical approach of having members of an autonomous swarm take in information from all surrounding members, the researchers opted for a simpler approach — likened to a smoke-filled room where only one person can see the exist, and nobody knows which person that is. "hat’s the counterintuitive part," Bhamla explains. "When only one person has the right information, averaging can wash out the signal. But if you follow one person at a time, and keep switching who that is, the right information can spread through the crowd."

Known as the Indecisive Swarm Algorithm, the control approach the team developed shows promise: in noisy conditions, when traditional algorithms break down, having the swarm members switch automatically between following a single neighbor or a guiding signal reduced the effort required to follow a desired path. "Our findings suggest that the same dynamics that make small animal groups unpredictable," Bhamla concludes, "may also offer new ways to control complex engineered systems."

The team's work has been published under open-access terms in the journal Science Advances.