Bottom-up Multi-agent Reinforcement Learning for Selective Cooperation

Aotani, Takumi; Kobayashi, Taisuke; Sugimoto, Kenji

doi:10.1109/smc.2018.00607

Cited by 2 publications

(1 citation statement)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, while fish form schools to protect themselves from predators, the dynamics of each individual is not necessarily uniform [15]. On the other hand, in the context of the swarm robotics, heterogeneity within the swarm can be found due to fluctuations in production processes [16,17] or by the intent of the operator of the swarm [18]. In order to address the aforementioned gap between the literature and the practice, Himo et al [19] recently proposed a guidance method for a flock containing agents not responding to the shepherd agent.…”

Section: Introductionmentioning

confidence: 99%

Shepherding algorithm for heterogeneous flock with model-based discrimination

2022

View full text Add to dashboard Cite

The problem of guiding a flock of agents to a destination by the repulsion forces exerted by a smaller number of external agents is called the shepherding problem. This problem has attracted attention due to its potential applications, including diverting birds away for preventing airplane accidents, recovering spilled oil in the ocean, and guiding a swarm of robots for mapping. Although there have been various studies on the shepherding problem, most of them place the uniformity assumption on the dynamics of agents to be guided. However, we can find various practical situations where this assumption does not necessarily hold. In this paper, we propose a shepherding method for a flock of agents consisting of normal agents to be guided and other variant agents. In this method, the shepherd discriminates normal and variant agents based on their behaviors' deviation from the one predicted by the potentially inaccurate model of the normal agents. As for the discrimination process, we propose two methods using static and dynamic thresholds. Our simulation results show that the proposed methods outperform a conventional method for various types of variant agents.

show abstract