Birds flock to forage and travel more efficiently, fish form schools to evade predators, and bees swarm to reproduce. Recent AI mimics natural group behavior with synchronized drones and robots for search-and-rescue and wildfire tracking. Yet, effectively managing and applying this form of AI, known as “swarm intelligence,” remains a significant challenge.
New Framework Mimics Nature to Enhance Robotic Swarm Intelligence
In a paper published in the Proceedings of the National Academy of Sciences, an international team of researchers presents a new framework aimed at advancing swarm intelligence by guiding flocking and swarming behaviors in ways that closely resemble those found in nature.
“One key hurdle in robotic swarms is decentralized control,” says Matan Yah Ben Zion, assistant professor at Radboud University.
“Fish, bees, and birds excel at this—they create complex formations and operate seamlessly without a central leader or commands. In contrast, artificial swarms lack that agility, and large-scale control remains out of reach.”
To address these issues, NYU scientists Mathias Casiulis and Stefano Martiniani developed geometric design principles for organizing self-propelled particles. Natural computation inspires these principles, as protons and electrons actively attract and repel to drive the formation of matter.
The Intrinsic Property Guiding Particle Motion
According to these rules, active particles that move in response to external forces possess an inherent characteristic that makes their paths curve—a property the researchers refer to as “curvity.”
“This curvature shapes swarm behavior, helping control flocking, flowing, or clustering,” says NYU’s Martiniani, assistant professor of physics, chemistry, and mathematics.
Experiments showed this curvature principle governs robot attraction—from pairs up to thousands. The team assigned each robot a positive or negative curvature that determined their interactions, like electric charge.
“This charge-like property, called ‘curvity,’ can be positive or negative and built into the robot’s design,” explains Ben Zion.
“Just like with electric charges, a robot’s curvity determines whether it’s drawn to other robots to form clusters or repelled in a way that leads to flocking behavior.”
From Microscopic Swimmers to Industrial Robots
Ben Zion, a former NYU student, said, “Identifying curvature as a design rule makes this approach adaptable to both large industrial robots and microscopic medical robots.”
“These rules use simple mechanics, making them easy to apply to robots,” adds Casiulis, a postdoc at NYU.
“This research shifts swarm management to materials science, offering clear design guidelines for future technologies.”
Read the original article on: Tech Xplore
Read more: Robot Handles Large Objects Like a Human After One Lesson