Utilizing embedded machine learning as a conversion algorithm for converting sensor readings into sensor values on a bare metal embedded device
Balder Grenness Klanderud · NORA - Norwegian Open Research Archives · 2020
Even though artificial intelligence (AI) has been frequently used for decades, the number of applications utilizing AI are rapidly increasing. The technology is advancing fast, and its use is spreading to different and new fields. In this thesis, the goal is to further expand the use and value of AI, by utilizing machine learning (ML) as a conversion algorithm for converting raw sensor readings, into the intended measured component (e.g. CO concentrations from a gas sensor). Using machine learning for this purpose is not new, however previous research on this topic have not produced machine learning models designed for use on the edge where the sensor resides. Thus, in this thesis, the main focus is placed on developing a model that can be implemented on a bare metal embedded device.