Self-organizing nervous systems for robot swarms.

Area of Science:

• Robotics
• Artificial Intelligence
• Complex Systems

Background:

• Swarm robotics typically relies on decentralized control.
• Existing architectures often struggle with dynamic reconfiguration and local coordination.
• Scalability, flexibility, and fault tolerance are key challenges in swarm systems.

Purpose of the Study:

• To introduce the self-organizing nervous system (SoNS) as a new robot swarm architecture.
• To demonstrate autonomous establishment, maintenance, and reconfiguration of dynamic multilevel hierarchies.
• To show coordinated sensing, actuation, and decision-making in a locally centralized manner.

Main Methods:

• Development of the SoNS architecture based on self-organized hierarchy.
• Implementation of dynamic multilevel system architectures for robot swarms.
• Proof-of-concept missions using a heterogeneous aerial-ground robot swarm.
• Physics-based simulation for scalability testing (up to 250 robots).

Main Results:

• Autonomous transformation of robot swarms into single or multiple SoNSs with dynamic hierarchies.
• Successful coordination of sensing, actuation, and decision-making.
• Advancement of state-of-the-art swarm robotics capabilities in missions like binary decision-making and search and rescue.
• Demonstrated scalability and fault tolerance in both simulation and real-world experiments.

Conclusions:

• The SoNS approach enables robots to autonomously create and manage complex, dynamic hierarchies.
• It achieves local centralization without compromising swarm benefits like scalability and fault tolerance.
• SoNS significantly enhances the capabilities and applicability of swarm robotics.