| The topic of robot navigation has been examined for many years from multiple points of view. Researchers in the field of Artificial Intelligence and Control Systems have proposed many solutions to specific problems involving system modeling, planning, optimization, stabilization, tracking, etc. Although Artificial Intelligence techniques have shown practical results over a range of robot navigation applications, they seem to fall behind in terms of mathematical formalism when compared to the theory developed in Control Systems. In this research we define a Multiresolutional architecture capable of accommodating AI representations and algorithms of deliberation along with Control systems techniques for reactive behavior. The architecture includes a formalism for Multiresolution system representation that accounts for implementability, consistency, controllability and stability properties of the overall system. The theoretical framework is applied, exemplified in a system capable of navigating autonomously through an environment. |
