A neural model of binocular saccade planning and vergence control

Abstract

The human visual system uses saccadic and vergence eye movements to foveate visual targets. To mimic this aspect of the biological visual system the PC/BC-DIM neural network is used as an omni-directional basis function network for learning and performing sensory-sensory and sensory-motor transformations without using any hardcoded geometric information. A hierarchical PC/BC-DIM network is used to learn a head-centred representation of visual targets by dividing the whole problem into independent subtasks. The learnt headcentred representation is then used to generate saccade and vergence motor commands. The performance of the proposed system is tested using the iCub humanoid robot simulator.

A neural model for eye–head–arm coordination

“…All experiments reported here were performed using the PC/BC-DIM hierarchical basis function neural network architecture proposed in [25][26][27][28][29]. Each level, or processing stage, in the hierarchy is implemented using the neural circuitry of three separate neural populations: the error, the prediction and the reconstruction neurons.…”

“…In [25] we built a three stage Predictive Coding/Biased Competition-Divisive Input Modulation (PC/BC-DIM) basis function network to control eye movements which we extended by adding one more PC/BC-DIM stage for coordinated eyes-head control [26] . In this article, we extend the network model proposed in [26] using another PC/BC-DIM stage to control arm movements for 3-D coordinated eyes-head-arm movements and to demonstrate the working of the proposed PC/BC-DIM basis function model.…”

“…The proposed model controls coordinated eyes-head-arm movements in 3-D space for direct and inverse transformations. Specifically, to perform a direct visuo-motor transformation the proposed network transforms retinotopic visual information together with binocular eyes position information into a head-centred representation as in [25]. The head-centred representation is combined with information about the head position into a bodycentred representation [26].…”

A Portable Bio-Inspired Architecture for Efficient Robotic Vergence Control

“…We call γ h and ν h the horizontal version and vergence angles; analogously, γ v and ν v are the vertical version and vergence angles. Assuming to know the projections of a 3D point in space on the left x L and the right x R camera image planes, commonly the version and vergence control are computed respectively by the average and by the difference of their coordinates (Samarawickrama and Sabatini 2007;Zhang and Phuan 2009;Muhammad and Spratling 2015):…”

“…Considering a binocular active visual system with a vergent stereo geometry, a common approach to guide binocular fovation is to select the target on the two retinas, and to compute the camera control as a combination of a conjugate version movement, to shift the binocular line of sight towards the object of attention, and a disconjugate vergence, to properly align the optical axes in depth (Enright 1998;Samarawickrama and Sabatini 2007;Zhang and Phuan 2009;Muhammad and Spratling 2015). From this perspective, dynamic vergence comes to be an instrumental resource not just for an expansion of the visual field, but mainly because at close distances the eyes' coordination ensures binocular fusion, providing a powerful and reliable source of information for visually guided behaviours.…”