SubCDM: Collective Decision-Making with a Swarm Subset

2025-08-01

AI Summary

To address the limitations of conventional swarm robotic collective decision-making—namely, mandatory participation of all agents, high resource overhead, and low task flexibility—this paper proposes a decentralized adaptive subgroup decision-making mechanism. The method dynamically constructs a minimal sufficient decision subgroup based solely on local sensory information, with subgroup size adaptively regulated by the joint influence of environmental features and consensus difficulty—thereby eliminating global communication and redundant computation. In large-scale simulations involving 100 robots, the approach achieves decision accuracy and full-swarm consensus while requiring only 23% of robots, on average, to participate in decision-making, significantly reducing communication and computational costs. Crucially, non-participating robots remain available for concurrent task execution, enhancing system concurrency and scalability. The core contribution is the first demonstration of dynamic, lightweight subgroup formation that guarantees consensus precision—breaking away from the traditional all-to-all participation paradigm.

Technology Category

• Intelligent Robots: Learning & Optimization for ROB
• Multiagent Systems: Distributed Problem Solving
• Humans and AI: Planning and Decision Support for Human-Machine Teams

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Fairness and ethical considerations in crowd work and in human-in-the-loop AI systems
• Systems and Infrastructure for Web, Mobile and WoT: Decentralized Web and Fediverse systems

Abstract

Collective decision-making is a key function of autonomous robot swarms, enabling them to reach a consensus on actions based on environmental features. Existing strategies require the participation of all robots in the decision-making process, which is resource-intensive and prevents the swarm from allocating the robots to any other tasks. We propose Subset-Based Collective Decision-Making (SubCDM), which enables decisions using only a swarm subset. The construction of the subset is dynamic and decentralized, relying solely on local information. Our method allows the swarm to adaptively determine the size of the subset for accurate decision-making, depending on the difficulty of reaching a consensus. Simulation results using one hundred robots show that our approach achieves accuracy comparable to using the entire swarm while reducing the number of robots required to perform collective decision-making, making it a resource-efficient solution for collective decision-making in swarm robotics.

Problem

• Enables consensus decisions using only a dynamic robot subset
• Reduces resource use by avoiding full swarm participation
• Adapts subset size based on decision difficulty for accuracy

Innovation

• Subset-Based Collective Decision-Making (SubCDM)
• Dynamic and decentralized subset construction
• Adaptive subset size for accurate decisions