A nervous system-inspired framework for deploying self-organizing robot swarms

Deploying robot teams could enable humans to complete various real-world tasks faster and more efficiently. For example, multiple cooperating robots could help quickly find and rescue survivors of natural disasters or monitor pollution across large geographic areas.

Researchers from the Free University of Brussels (ULB) have developed a new swarm architecture inspired by the human nervous system that could improve cooperation between robots within teams. The proposed approach, described in an article published in Scientific roboticsallows robots to self-organize into sub-swarms, improving their coordination as they sense their surroundings, move, and plan their next steps to accomplish a mission.

“Over the past two decades, swarm robotics research has demonstrated a wide range of powerful collective behaviors that do not require any central coordinating entity or process,” said Dr. Mary Katherine Heinrich, co-first author of the study. article and postdoctoral researcher at the University of Sydney. ULB’s IRIDIA artificial intelligence laboratory, reported Tech Xplore.

“Despite such progress, robot swarms still struggle to move from laboratory experiments to real-world applications. Indeed, from the perspective of many application domains, self-organization also has significant drawbacks .”

Achieving self-organization in robot swarms is an ambitious research goal. This is because even though desired behaviors occur at the group level, robots are programmed individually, making analytical design of swarm behaviors extremely difficult.

“Developing new swarm behaviors is a lengthy process of trial and error and, once new swarm behaviors have been developed, they cannot be easily modified or combined,” explained Professor Marco Dorigo , lead author of the article and director. from the ULB IRIDIA artificial intelligence laboratory.

“In our paper, we address this challenge by combining aspects of centralized control with aspects of self-organized control, to attempt to exploit the advantages of both in a unified system.”

The new framework works by building and rebuilding self-organizing hierarchies. In other words, it allows robots in a team to self-organize into an ad hoc dynamic control network, called a self-organizing nervous system (SoNS).

As this network forms, robots temporarily and interchangeably occupy specific positions within a leadership hierarchy, similar to that of the human nervous system. The highest position in this hierarchy is that of the “brains”, who guides and oversees the group’s efforts during a mission.

“In a SoNS control network, each robot communicates only with its direct neighbors, to avoid the type of bottleneck that would occur at the communications center in a fully centralized system,” Dr. Heinrich said.

“Depending on task specifications and system constraints, sensor information may be fused when transmitted upstream, control information may not be fused when transmitted downstream, and the balance between individual and collective behavior can be actively managed.”

The SoNS swarm architecture acts as a sort of “middleware” for robots, allowing each robot to organize itself into dynamic hierarchies. In the resulting network, robots can leverage their capabilities as a team to best accomplish a given task.

“Through SoNS, robots can coordinate their collective sensing, actuation, and decision-making activities centrally locally, without sacrificing the scalability, flexibility, and fault-tolerance benefits normally associated with self- organization,” explained Professor Dorigo.

“In other words, the SoNS architecture effectively allows a swarm of robots to be programmed as if it were a single robot, which we believe can greatly improve the transferability of robot swarms from laboratory to real-world applications.”

Dr Heinrich, Professor Dorigo and their colleagues tested the proposed framework in swarm simulations with up to 250 aerial and ground robots and in proof-of-concept experiments with real robots. The results of these tests were very promising, because their approach made it possible to effectively coordinate the actions of many robots.

In the future, the team may conduct further testing to evaluate its framework in a wider range of scenarios. In the meantime, they plan to further improve the proposed swarm architecture, in order to facilitate its future deployment on real robots.

“One of the most exciting directions for future research with SoNS is the development of more advanced SoNS brains and more advanced hierarchical computing, enabling for example online learning or autonomous mission planning capabilities,” added Dr Heinrich and Professor Dorigo.

© 2024 Science X Network