Gaze Control for a Two-Eyed Robot Head

Abstract

As a first step towards building a two-eyed active vision system, gaze control is discussed in this paper. Instead of using separate sub controllers for each of the subfunctions, loosely corresponding to saccade, pursuit, vergence, VOR(vestibulo-ocular reflex), and OKR(opto-kinetic reflex), that most previous work has done, a potentially parallel gaze controller is proposed whose structure is supervising-planning-adaptation. Based on simulation, the co-operations and interactions in gaze control and the consequence of delays due to image processing and the local controllers are discussed.