This project addresses the difficult problem of detecting toxic gases. Sensors have been made from semiconductors which could make it much cheaper and easier to detect such gases, but it is difficult and expensive to do the calculations needed to interpret what the sensors are measuring. In phase I of this project, the company successfully proved that artificial neural networks (ANN's) can be used to reduce the cost of interpretation by orders of magnitude, for detecting three test gases. Phase II will extend this further, by trying to detect carbon monoxide (CO) -- a very important pollutant -- under a variety of conditions, with interference from humidity and other gases. It will attempt to measure CO concentration levels accurately, in real- time.
