Uninorm-based neural network and its application for control of mobile robots

Abstract

We suggest the model of recurrent neural network based on the uninorm aggregators and apply it for control of mobile robots. The learning process in the network is governed by the changes of the values neutral elements. The mobile robots in the group are considered as mobile neurons such that their mobility is defined with respect to their internal states. For the suggested model we construct non-monotonic generator function that, however, preserves its monotonicity in the algebraic structure defined by the uninorm and absorbing norm aggregators. The model was verified by numerical simulations and by the trials with the mobile robots.