Curvity is the new way to control the drone swarms.
"Illustration of robots advancing artificial swarm intelligence inspired by the collective behaviors of birds, fish, and bees."(Rude Baguette, “Curvity Controls Robot Interactions”: Researchers Develop New Framework Allowing Robotic Swarms to Mimic Bird and Fish Group Behaviors)
Drone swarms could be the tool that can make almost everything. The idea is that drones can cooperate and connect their computing capacity when they must make complex decisions. Then the swarm can share missions to individual drones. When we think about robot swarms that can mimic things like bird swarms, those robots must know their position in the swarm. The answer to that question can be curvature or curvity. Rude Baguette tells the next things about curvity.
“A key innovation of the study is the introduction of a new quantity termed “curvity.” This intrinsic charge-like quality allows a robot to curve in response to external forces, guiding its interactions with fellow robots. Each robot is assigned a positive or negative curvity value, which determines its behavior within the swarm. This innovative approach allows for the collective behavior of the swarm to be controlled, whether it involves flocking, flowing, or clustering.” (Rude Baguette, “Curvity Controls Robot Interactions”: Researchers Develop New Framework Allowing Robotic Swarms to Mimic Bird and Fish Group Behaviors)
Assistant Professor Stefano Martiniani from New York University emphasized the potential of this approach: “This curvature drives the collective behavior of the swarm, potentially controlling whether the swarm flocks, flows, or clusters.” Such a model transforms the challenge of controlling swarms from complex programming into a material science issue, opening new avenues for research and application.” (Rude Baguette, “Curvity Controls Robot Interactions”: Researchers Develop New Framework Allowing Robotic Swarms to Mimic Bird and Fish Group Behaviors)
Sharing the system into substructures that mimic galactic groups makes it easier to create programs for robots. Each layer or level acts as one robot.
The system can involve the next type of information.
“You are a robot of robot group 9. Robots 1 and 2 must be ahead of you. And robot 4 must be on the left side of you. Together, you are Robot Swarm 2, and Swarms 1 must be on your side, and Swarms 3 and 4 must be behind you. Together, your place is in layer 3. That layer must be between layer 2 and 4.” Then the system has other algorithms that tell what to do if one drone is lost.
The key question is how the robot can determine its position within the swarm. In this case, the modular structure is the solution. There can be multiple data levels, which makes the data structure operate as a mosaic. Or, rather, we can say that the structure mimics the universe. There are individual drones that are galaxies in galactic clusters. Then those drone clusters form a local cluster. Then, local clusters form a super cluster. And then the superclusters form a global cluster. Those clusters are the drone swarms. The ability to share the groups into subgroups makes it easier to program those drones.
The idea is that the drones can form three or four drone groups. The individual drone knows. What drones are behind it, and what drones are ahead of it? The drone group acts like one drone. And then it must know what drone group is ahead and behind it. Then the drone group must know. Whether it should be above or below some of the other drones. The system can make that quite easily. It can share the drone swarm into layers. And those drones know their own layer. And then they know what drones should be below and what should be above it. Those drones can operate in squares. That can look like randomly changing forms.
https://www.rudebaguette.com/en/2025/09/curvity-controls-robot-interactions-researchers-develop-new-framework-allowing-robotic-swarms-to-mimic-bird-and-fish-group-behaviors/