Current work refers to the effort of our group to extend abilities and functionalities of embodied agents closely connected in a communication network. The capability to develop a collective and social intelligence in poor communication conditions, demands for an effective sharing of knowledge among the agents. The key is to optimize the communication and the information/knowledge flow from the source towards the target nodes. Alarm management, cooperative target recognition, shared learning, imitation learning all take advantage of an effective knowledge sharing management. Being the timing controlled by the external events, the challenge becomes to define the reaching limit of information and knowledge: This limit is critical for the best reaction of the system to the environmental solicitations. The Knowledge Horizon concept is the consequence: A spatial region and its relevant set of nodes/agents that is the limit of data propagation, determined by the communication channel, by the timing of external events and by the payload to be transferred. In this first stage of the work we propose a shortest path approach that, given an estimate of the delays on each arc of the communication network and an estimate of the dynamics of an external event, can determine both the set of nodes in the knowledge horizon and a multicast strategy to efficiently forward the information to it. © IFAC.
|Titolo:||Knowledge management and routing in embodied agents networks|
|Data di pubblicazione:||2013|
|Appare nelle tipologie:||4.1 Contributo in Atti di convegno|