SBIR Phase II: 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 (SBIR) Phase II Project proposes the construction of a miniature object recognition and color segmentation system on a chip. This chip will be tuned to recognize various predefined targets in natural environments. The chip will use an object recognition model, color histogramming, originally derived from research in cognitive neuroscience. Taking advantage of recent advances in Neuromorphic Engineering, the company will implement the basic sensing and computational elements directly in silicon using mixed analog/digital processing. In contrast, implementing the same model or algorithm with conventional microprocessor technology would require that the basic computations be simulated as an intermediate step. The removal of this intermediate step will result in an intelligent sensor with dramatically lower cost, smaller volume, and reduced power usage-achievements not possible using competing microprocessor-based technology.The applications for this technology include intelligent toys and prosthetic devices. A toy might be made to recognize, and therefore be able to respond to, the presence of another toy or specially designed environment. More advanced and elaborated versions of the chip might be used as an aid to the blind by assisting them in finding standardized (i.e. specially colored) objects. For example, a blind person might be assisted in localizing a coffee mug, distinguishing between two similar items of clothing differing only in color, or finding a standardized 'EXIT' sign in a building. The broader impact of this technology is that it will help bridge the gap between the natural, unstructured environment and computing technology.