Multi-robot Cooperation Based on Hierarchical Reinforcement Learning

Cheng, Xiaobei; Shen, Jing; Liu, Hai-Bo; Gu, Guochang

doi:10.1007/978-3-540-72588-6_12

Cited by 17 publications

(12 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It learns behavior for sub-tasks given a hand-designed hierarchy. Many extensions to this work have been created since, including a multi-agent variant (Cheng et al, 2007) and a version that learns the hierarchy in addition to the sub-tasks (Hengst, 2002 …”

Section: Hierarchical Reinforcement Learningmentioning

confidence: 99%

Evolving multimodal behavior through modular multiobjective neuroevolution

Schrum

2015

SIGEVOlution

View full text Add to dashboard Cite

Intelligent organisms do not simply perform one task, but exhibit multiple distinct modes of behavior. For instance, humans can swim, climb, write, solve problems, and play sports. To be fully autonomous and robust, it would be advantageous for artificial agents, both in physical and virtual worlds, to exhibit a similar diversity of behaviors. Artificial evolution, in particular neuroevolution [3, 4], is known to be capable of discovering complex agent behavior. This dissertation expands on existing neuroevolution methods, specifically NEAT (Neuro-Evolution of Augmenting Topologies [7]), to make the discovery of multiple modes of behavior possible. More specifically, it proposes four extensions: (1) multiobjective evolution, (2) sensors that are split up according to context, (3) modular neural network structures, and (4) fitness-based shaping. All of these technical contributions are incorporated into the software framework of Modular Multiobjective NEAT (MM-NEAT), which can be downloaded here.

show abstract

Section: Hierarchical Reinforcement Learningmentioning

confidence: 99%

Evolving multimodal behavior through modular multiobjective neuroevolution

Schrum

2015

SIGEVOlution

View full text Add to dashboard Cite

show abstract

“…Simultaneously, agent 4 gains object 3 from agent 2 and agent 3 gains object 6 from agent 2. The final sets particular agents negotiated to collect are defined as follows: Finally, optimal solutions that are time-ordered sequences of objects the given agent planned to collect are: 16,6, 10} S * 4 = {7, 9, 3} A graphical interpretation of the transitions of agents resulting from this solution is presented in Fig. 5.…”

Section: Single Stage Case Studymentioning

confidence: 99%

“…On the other hand, there are a lot of challenges that must be met in order to design effective and robust systems that are able to solve problems or execute tasks. These challenges were discussed in [28] and it is enough to point out the problems like coordination [6,15,33], task division [11,14,30] etc. The potential advantages of MAS were quickly noticed by researchers who deal with problems related to Robotics.…”

Section: Introductionmentioning

confidence: 99%

Handling uncertainty of resource division in multi agent system using game against nature

Skrzypczyk¹,

Mellado²

2014

Archives of Control Sciences

View full text Add to dashboard Cite

. Knowledge of the environment represented in the system is uncertain, incomplete and distributed among the individual agents that have both limited sensing and communication abilities. The pick-up-and-collection problem is considered in order to illustrate the idea presented. In this paper a framework for cooperative task assignment to individual agents is discussed. The process of negotiating access to common resources by intercommunicating agents is modeled and solved as a game against Nature. The working of the proposed system was verified by multiple simulations. Selected, exemplary simulations are presented in the paper to illustrate the approach discussed.

show abstract

“…Therefore calculating the agent reliability as the average given by (14) seems to be valid approach. Finally, the reliability of the agent at the stage of the process can be evaluated as an exponential moving average with a time window of length .…”

Section: Agent Reliability Assessmentmentioning

confidence: 99%

“…On the other hand, there are a lot of challenges that must be met in order to design effective and robust systems that are able to solve problems or execute tasks. These challenges were discussed in [12] and it is enough to point out the problems like coordination [2], [4], [5], [13], [14] task division [3], [7], [10], [17], cooperation [1], [7], [8], [15] etc. The potential advantages of MAS were quickly noticed by researchers who deal with problems related to Robotics.…”

Section: Introductionmentioning

confidence: 99%

Information Fusion in Multi-agent System Based on Reliability Criterion

Mellado

Skrzypczyk

2013

Vision Based Systemsfor UAV Applications

View full text Add to dashboard Cite

SpringerMellado Arteche, M.; Skrzypczyk, K. (2013). Information fusion in multi-agent system based on reliability criterion. En Vision Based Systemsfor UAV Applications. Springer. 207-217. doi:10.1007/978-3-319-00369-6_13. Information Fusion in Multi-agent System Based on Reliability CriterionMartin Mellado and Krzysztof Skrzypczyk * Abstract. The paper addresses the problem of information fusion in Multi-Agent System. Since the knowledge of the process state is distributed between agents, the efficiency of the task performance depends on a proper information fusion technique applied to the agents. In this paper we study the case in which each agent has its own sensing device and is able to collect information with some certainty. Since the same information can be detected by multiple agents, the global certainty about the given fact derives from the fusion of information exchanged by interconnecting agents. The key issue in the method proposed, is an assumption that each agent is able to asses its own reliability during the task performance. The method is illustrated by the pick-up-and-collection task example. The effectiveness of the method proposed is evaluated using relevant simulation experiments.

show abstract

Multi-robot Cooperation Based on Hierarchical Reinforcement Learning

Cited by 17 publications

References 8 publications

Evolving multimodal behavior through modular multiobjective neuroevolution

Evolving multimodal behavior through modular multiobjective neuroevolution

Handling uncertainty of resource division in multi agent system using game against nature

Information Fusion in Multi-agent System Based on Reliability Criterion

Contact Info

Product

Resources

About