Convolutional network for vergence control

We present a biologically-inspired model for the one-shot vergence control of a robotic head, which has been used for an investigation of two vergence control networks. Both networks do not work with explicitly computed disparity, but extract the vergence control signal from the postprocessed response of a population of disparity tuned complex cells, the actual gaze direction and the actual vergence angle. Training and evaluation of the networks are also discussed.