Embedding visual routines in AnaFocus' Eye-RIS Vision Systems for closing the perception to action loop in roving robots

摘要

The purpose of the current paper is to describe how different visual routines can be developed and embedded in the AnaFocus' Eye-RIS Vision System on Chip (VSoC) to close the perception to action loop within the roving robots developed under the framework of SPARK II European project. The Eye-RIS Vision System on Chip employs a bio-inspired architecture where image acquisition and processing are truly intermingled and the processing itself is carried out in two steps. At the first step, processing is fully parallel owing to the concourse of dedicated circuit structures which are integrated close to the sensors. At the second step, processing is realized on digitally-coded information data by means of digital processors. All these capabilities make the Eye-RIS VSoC very suitable for the integration within small robots in general, and within the robots developed by the SPARK II project in particular. These systems provide with image-processing capabilities and speed comparable to high-end conventional vision systems without the need for high-density image memory and intensive digital processing. As far as perception is concerned, current perceptual schemes are often based on information derived from visual routines. Since real world images are quite complex to be processed for perceptual needs with traditional approaches, more computationally feasible algorithms are required to extract the desired features from the scene in real time, to efficiently proceed with the consequent action. In this paper the development of such algorithms and their implementation taking full advantage of the sensing-processing capabilities of the Eye-RIS VSoC are described