Compiling CNNs with Cain: focal-plane processing for robot navigation

Focal-Plane and Multiple Chip VLSI Approaches to CNNs

In this paper, three alternative VLSI analog implementations of CNNs are described, which have been devised to perform image processing and vision tasks: a programmable low-power CNN with embedded photosensors, a compact fixed-template CNN based on unipolar current-mode signals, and basic CMOS circuits to implement an extended CNN model using spikes. The first two VLSI approaches are intended f...

Focal-Plane Optical Flow Computation by Foveated CNNs

Optical flow computation is instrumental in robot guidance. Optoelectronic smart-pixel sensors for such computation may be realized on a single chip, by making use of a suitable Cellular Neural Network architecture defined on a log-polar space-variant grid. Simulations confirm validity of the filtering system, and possible realizable structures are discussed.

Robot Navigation in the Plane � �

Architectures for focal plane image processing

Architectures for focal plane image processing are discussed. On-chip image preprocessing for solid-state imagers using analog CCD circuits is described for low, medium, and high density detector arrays. A spatially parallel architecture for low density, high throughput applica­ tions is described. For sparse illumination or event detection, a content· addressable architecture is proposed. A ne...

FOCAL POINT AND FOCAL K-PLANE

This paper deals with the basic notions of k-tautimmersions . These notions come from two special cases; that is, tight and taut immersions. Tight and taut based on high and distance functions respectively and their basic notions are normal bundle, endpoint map, focal point, critical normal. We generalize hight and distance functions to cylindrical function and define basic notions of k-taut ...