Prediction in Gaze and Saccade Control

摘要

Animate vision system, biological or robotic, employ gaze control systems to acquire, fixate, and stabilize images. Gaze control and other sensorimotor skills, are complicated by two main problems: time delay and the interaction of control subsystems. One solution is open-loop control. The other, explored here, is control with prediction. The goal is to build robust gaze control behavior from cooperating lower level visual reflexes. Solutions are explored through simulation incorporating ten primitive control capabilities, more or less comparable to subsystems in primate gaze and head control. Versions of several of the subsystems have been implemented on a binocular robot. Signal synthesis adaptive control with Smith prediction is the basic paradigm, implemented with kinematic simulation of the agent and optimal filtering to predict world state. The task of rapid gaze shift illustrates several issues. Predictive methods appear useful for artificial gaze control systems and often produce results consistent with physiological data. Keywords: Dynamic system estimation.