Learning Grouping and Anti-predator Behaviors for Multi-agent Systems

Morihiro, Koichiro; Nishimura, Haruhiko; Isokawa, Teijiro; Matsui, Nobuyuki

doi:10.1007/978-3-540-85565-1_53

Cited by 11 publications

(10 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…DEVS capabilities in the field of two practical examples, namely, a model of unmanned aerial vehicles and a model of an automated systemare presented.The described ________________________________________________________________________________________________ 26 defence models have been previously presented and well-received at the Baltic Defence Research and Technology Conference, Riga,September [10][11]2009. Although additional experiments might be further required, computational results conducted so far clearly show that the V-DEVS approach is suitable for the development and analysis of defence models.Future along this line of research includes improvement and validation of the described simulation algorithms and models.…”

Section: Discussionmentioning

confidence: 99%

“…The reinforcement learningapproach, namely Q-Learning,is used for the search of a nearly optimal solution in the patrolling area. The reinforcement learning is a machine learning approach in which transition rules are learned through the experience of agents in their environment.Q-learning [10]is known as the best understood reinforcement learning technique used for maximizing the sum of the rewards received.In Q-Learning, the learning process consists of acquiring a state t s , deciding an action t a , receiving a reward r from an environment, and updating Q-value ( ) , t t Q s a [11] calculated by the following equation:…”

Section: A Model Of Unmanned Aerial Vehiclesmentioning

confidence: 99%

See 1 more Smart Citation

Creating Defence Models Using V-DEVS Framework

Lektauers¹,

Merkuryev²

2010

Scientific Journal of Riga Technical University. Computer Sciences

View full text Add to dashboard Cite

The aim of this paper is to present a development technique for virtual/constructive simulation of defence applications. The approach relies on the use of V-DEVS methodology providing a modular, hierarchical modelling structure and enabling to effectively synchronize simulation, visualization and user interaction. The practical importance and application possibilities of the research results are demonstrated by developing simulation models of unmanned aerial vehicles and a flexible manufacturing system producing military SUVs.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: A Model Of Unmanned Aerial Vehiclesmentioning

confidence: 99%

Creating Defence Models Using V-DEVS Framework

Lektauers¹,

Merkuryev²

2010

Scientific Journal of Riga Technical University. Computer Sciences

View full text Add to dashboard Cite

show abstract

“…They find that the swarming among prey is influenced by how predators attack. While Reynolds (1999) proposes static rules leading to swarming, Morihiro et al (2008) have their agents learn to do so by encoding these rules in the reward of an RL algorithm. Further, Sunehag et al (2019) find the emergence of flocking and symbiosis with rewards shaping by independent MARL in simulated multiple-species ecosystems.…”

Section: Further Emergent Behaviormentioning

confidence: 99%

A Sustainable Ecosystem through Emergent Cooperation in Multi-Agent Reinforcement Learning

Ritz¹,

Ratke²,

Phan³

et al. 2021

The 2021 Conference on Artificial Life

View full text Add to dashboard Cite

This paper considers sustainable and cooperative behavior in multi-agent systems. In the proposed predator-prey simulation, multiple selfish predators can learn to act sustainably by maintaining a herd of reproducing prey and further hunt cooperatively for long term benefit. Since the predators face starvation pressure, the scenario can also turn in a tragedy of the commons if selfish individuals decide to greedily hunt down the prey population before their conspecifics do, ultimately leading to extinction of prey and predators. This paper uses Multi-Agent Reinforcement Learning to overcome a collapse of the simulated ecosystem, analyzes the impact factors over multiple dimensions and proposes suitable metrics. We show that up to three predators are able to learn sustainable behavior in form of collective herding under starvation pressure. Complex cooperation in form of group hunting emerges between the predators as their speed is handicapped and the prey is given more degrees of freedom to escape. The implementation of environment and reinforcement learning pipeline is available online. 1

show abstract

“…Supplementary forces can be introduced, which repel an individual from an enemy or from obstacles, for example. To overcome these static rules definitions, Morihiro et al (2008) used Reinforcement Learning, particularly Q-Learning, to train agents to follow these rules. In their model the agents iteratively learn while at every time step an agent i only considers one other agent j.…”

Section: Swarm Behaviormentioning

confidence: 99%

“…The question arises whether flocking behavior occurs in a scenario with such properties where agents solely try to maximize their survival time. In contrast to Morihiro et al (2008), we pursue a scenario in which agents are trained with reinforcement learning solely on the objective to survive, without explicitly enforcing swarm behavior. Additionally, we demonstrate that SELFish also works for a continuous action space of the agents.…”

Section: Swarm Behaviormentioning

confidence: 99%

Emergent Escape-based Flocking behavior using Multi-Agent Reinforcement Learning

Hahn

Phan

Gabor

et al. 2019

The 2019 Conference on Artificial Life

View full text Add to dashboard Cite

In nature, flocking or swarm behavior is observed in many species as it has beneficial properties like reducing the probability of being caught by a predator. In this paper, we propose SELFish (Swarm Emergent Learning Fish), an approach with multiple autonomous agents which can freely move in a continuous space with the objective to avoid being caught by a present predator. The predator has the property that it might get distracted by multiple possible preys in its vicinity. We show that this property in interaction with self-interested agents which are trained with reinforcement learning to solely survive as long as possible leads to flocking behavior similar to Boids, a common simulation for flocking behavior. Furthermore we present interesting insights in the swarming behavior and in the process of agents being caught in our modeled environment.

show abstract

Learning Grouping and Anti-predator Behaviors for Multi-agent Systems

Cited by 11 publications

References 9 publications

Creating Defence Models Using V-DEVS Framework

Creating Defence Models Using V-DEVS Framework

A Sustainable Ecosystem through Emergent Cooperation in Multi-Agent Reinforcement Learning

Emergent Escape-based Flocking behavior using Multi-Agent Reinforcement Learning

Contact Info

Product

Resources

About