Collision avoidance using an egocentric memory of proximity

Abstract

This paper describes a formal model of a dynamic and egocentric memory used for predicting the evolution of a local function of proximity. This model is represented by a polar function able to trig the reactive behaviors (reflex actions for collision avoidance) of a planar mobile robot. It also addresses the implementation of this model with an artificial neural network first on a simulator that has been developed in order to test this model and also on a real holonomic autonomous robot.