Autonomous Agents in Dynamic Environment: A Necessary Volatility of the Knowledge
Résumé
Completely autonomous vehicles should allow to decrease greatly the number of accident victims, and should allow gains in terms of performance and economy. Interactions among the different vehicles allowing them to choose the best path, the best behaviour is one of the main challenges. We propose in this paper a model of volatile knowledge dedicated to mobile agents on a traffic network. This model of knowledge and the principles of interactions allow to propagate new knowledge with a limited number of messages. For that, a degradation coefficient of the knowledge is proposed. The principles have been validated by a simulation with software agents, and by a real application on mobile robots acting like autonomous vehicles.