Coordination patterns in Multi-Agent Worlds applications to computer vision and robotics

Abstract

Presents a new paradigm in the field of Multi-Agents Worlds. The coordination patterns (the PACO project), in which the concepts of autonomous agents, multi agent systems, emergent functionality and complex dynamics meet. The originality of the approach is mainly due to the definition and the control of a perception scope and of a communication scope that guide the interaction between agent themselves and between the agents and the environment. The second originality of the approach is the explicitation of a decision filter that experimentally enables the entire system to faster stabilize into an equilibrium. This research has been initialized at LIFIA and then enhanced at VUB/AI-Lab, both through image analysis studies. From the beginning, however, several applications in different fields were envisioned. The author is applying the coordination patterns to the path-planning and the execution control problems in robotics, when spatially taking into account the temporal constraints.