Intelligent Agents: The Measure of their Intelligence

摘要

This paper presents an original attempt for testing the level of autonomy of intelligent physical agents. After a short review of the architectural problematic of designing such artefacts, we try to situate the place of uncertainty modelling in mobile robotic and complex control systems. We then briefly describe a design for giving to an industrial control problem the highest level of autonomy for performing its tasks. We give some hints how to extend these tests on physical agents exhibiting adaptive behaviours. Future works for applying this methodology will be illustrated.