| The goal of this thesis is to develop a binocular controller for artificial vision systems, based on principles used in biological binocular control. The controller is a symmetrical, two-sided system utilizing feedback across the midline. In spite of its simple topology, the controller allows for the integration of various aspects of eye movements, which are often modelled as independent systems. In addition, the controller topology naturally lends itself to sensory-fusion.;The controller was applied to an artificial vision system and its 3-D tracking performance was evaluated. Eye (camera) movements consist of an appropriate mix of slow (tracking) and rapid (reorienting) motions. This is the first implementation of an artificial vision control system which: (1) Controls two eyes independently, in 2-D, for target motion in one plane. (2) Provides integrated conjugate and vergence coordination. (3) Provides integrated slow and fast operating modalities. (4) Has the intrinsic ability to hold gaze in the absence of a target.;The controller is also well suited for extensions to coordinate multiple platforms (e.g. eyes on mobile head) using a conceptually simple motor-error control scheme. Its sensory-fusion properties make the controller receptive to the inclusion of a VOR system, essential in mobile robots. |