"Probabilistic Modelling in Collective Autonomous Robotics: From Collective Manipulation and Exploration Case Studies towards an Unified Methodology"

11/21/2000: [time not recorded]
[location not recorded]

Abstract: In this talk, I will present a set of experiments which differ in the collective task performed (collective manipulation, collective exploration) and in the type of the robotic platform used. The common signatures of all the experiments can be summarized as follows: the robots are fully autonomous, are endowed with minimal individual capabilities, and the control of the group is fully scalable and distributed (Swarm Intelligence approach). In order to systematically study and understand the key parameters of the robotic system (robot control, hardware morphology, sensorial features, number of robots, and so on) playing a crucial role in the team performance, we developed and applied probabilistic modelling methods to several different case studies in two classes of collective experiments (manipulation and exploration). I will present results, advantages, and limitations of the current methods and the future challenges to be faced for developing an unified methodology at least for each class of experiment.

Last updated: Mon Jun 19 17:44:06 2006