Controlling noisy herds: Temporal network restructuring improves control of indecisive collectives

Area of Science:

• Control theory
• Network science
• Collective behavior

Background:

• Controlling stochastic temporal networks is challenging due to unpredictable dynamics.
• Real-world systems often exhibit stochasticity, complicating control efforts.
• Biological controllers offer insights into managing complex, dynamic systems.

Purpose of the Study:

• Investigate control mechanisms for stochastic temporal networks using sheepdog trials as a model.
• Analyze how biological controllers manage unpredictable groups.
• Develop and evaluate novel algorithms for controlling noisy swarms.

Main Methods:

• Modeled sheep behavior using a stochastic choice model.
• Analyzed sheepdog trial data to understand control strategies.
• Developed the Indecisive Swarm Algorithm (ISA) and compared it with Averaging-Based Swarm Algorithm (ASA) and Leader-Follower Swarm Algorithm (LFSA).

Main Results:

• Trained dogs exploit sheep indecisiveness as a control mechanism for herding and splitting.
• The Indecisive Swarm Algorithm (ISA) minimizes control energy in trajectory-following tasks.
• ISA outperforms standard algorithms, especially under noisy conditions.

Conclusions:

• Indecisiveness in collective behavior can be leveraged as a control tool.
• The ISA provides an effective framework for managing stochastic temporal networks.
• Findings have applications in swarm robotics, animal collectives, and opinion dynamics.