Artificial cognition for autonomous planar vehicles: modelling collision avoidance and collective manoeuvre

Abstract

A hierarchical cognitive robotics model for a team of unattended robotic ground vehicles (UGVs) is proposed. The first level rigorously defines conflict resolution for a couple of UGVs, using dynamical games on SE(2)-groups of plane motion. The second level extends it to n UGVs, using Nash-equilibrium approach. The third provides adaptive guidance for several groups of UGVs. The fourth, collective manoeuvre level, proposes a combination of an attractor neural model and a fuzzy-neural 'supervisor', to perform an adaptive path definition and waypoints selection, as well as chaos control. The fifth, cognitive level, performs overall mission planning/feedback control.