Developing a spectral pipeline using open source software and low-cost hardware for material identification

The ability to access, design and create low cost sensors capable of returning scientifically useful data has led to an exponential increase in citizen science, education and environmental monitoring groups. Low-cost spectroscopy is one such application and mobile phone camera-based instruments have been used in pollution monitoring, medical applications in developing countries and vegetation analysis. Can such an instrument be developed and tested to assist with automated detection of materials, possibly from space? We tested two spectrometer designs inside a two unit (2U) cubesat frame against a series of materials exhibiting phenomenology in the visible/near infrared (Vis/NIR) portion of the spectrum and vegetation groups. This was conducted in order to determine whether open source designs were capable of discriminating against similar materials, such as types of vegetation or types of iron-rich minerals. A spectral pipeline was created using open source programming software that was capable of converting raw sensor data into spectra, comparing samples of interest against a spectral library and returning an identification result with a confidence interval. We found that low-cost hardware sensitive to NIR and freely available software were able to identify types of materials in the study set, enabling applications in citizen science, education and outreach or even low-cost near-space research.