Methodologies and Tools for the Emergence of Cooperation in Biorobotics.

One of the main purposes of the Ph.D. activities was the investigation of the swarm aspects in order to formulate new strategies for the emergence of cooperation within a colony of robots. The proposed idea was to furnish each robot with identical cognitive architectures, applying them in a reward-based learning mechanism leading to the emergence of cooperation capabilities and perform tasks where one robot alone cannot succeed. The selected technique to induce the emergence of cooperation was based on a selection of specific sub-set among the available behaviours. During the learning phase, each robot can evolve its knowledge in order to specialize its behaviours to acquire a specific role in the environment challenge. The aim is to demonstrate how a group of robots, each one equipped with an independent control system, thanks to the flexibility of the architecture can learn cooperative behaviours through the specialization and organization of several activities and roles to reach a common intent. To permit the investigation of these new cooperation methodologies, one of the important aims was the development of new tools and platform for multi-robot applications. The thesis reports the methodology, the implemented tools and interesting applications which underline the potentialities of this emergent approach to cooperative biorobotics.

Grantor dc:publisher

Università degli studi di Catania

Year dc:date

2012

Author dc:creator

• VITANZA, ALESSANDRA

Contributors dc:contributor

• ARENA, Paolo Pietro
• FORTUNA, Luigi

• Swarm Intelligence,collaborative algorithms,labor division,swarm robotic applications,task partitioning,bio-robotics,spike-frequency adaptation,STDP learning,robotic framework,dynamic simulator

Statement dc:rights

• info:eu-repo/semantics/openAccess
• license:PUBBLICO - Pubblico con Copyright
• license uri:iris.PUB02

Language dc:language

eng

Handle dc:identifier

https://hdl.handle.net/20.500.11769/587725

OAI identifier oai:identifier

oai:www.iris.unict.it:20.500.11769/587725