SBIR Phase I: Neuromorphic Color Sensor for Object and Place Recognition

点击翻译按钮获取中文摘要

This Small Business Innovation Research Phase I project from Iguana Robotics, Inc. proposes the construction of a single chip, color aVLSI/dVLSI Neuromorphic sensor (camera) with onboard segmentation and object and place recognition capability. For each pixel, this chip will compute the transformation to hue and saturation values in the focal plane, perform class assignment based on color, and perform histogramming based on the class assignment. This chip will be part of a symbiotic system. In addition to the chip, Iguana Robotics will design special target color and texture coded patterns. By designing the targets to be easy to recognize, the investigators will achieve a very high recognition rate. The utility of this device is that it will (1) be capable of associating symbolic tags with objects in natural environment; (2) estimate the position of objects; (3) track and find human face and hands in an image as a pre-processing step in HCI applications; and (4) provide a technique for place recognition for personal robots. In addition, this chip that will be (1) very low power (1mw); (2) very low cost; and (3) very small and compact. These features will facilitate the wide spread use of this object in cost and power sensitive applications.Iguana Robotics proffers low-cost, compact neuromorphic color sensor for object and place recognition that can provide real-time object and place recognition for diverse areas of application such as entertainment, education, industry, interfaces, and banking. The research project has the potential to make a significant impact on the discovery and understanding of automation in gesture recognition and to advance the state-of-the-art in on-chip computer vision hardware.