Blog     Github

Overview

ScanSpectrum is a line of low-cost and portable spectrometers. We aim for both  affordability and high build quality by combining careful design with the use of cheap yet powerful commercial-off-the-shelf components.

Our inspiration for building ScanSpectrum stems from the scarcity of publicly available agricultural and medical data, particularly in the developing world. While there are many reasons for this, two pain points we wish to address are (1) the high prices of desktop lab instruments for health assessment, and (2) the logistical cost of ferrying samples between the field and the lab. ScanSpectrum aims to alleviate both problems by making sensing less expensive and eliminating most of the logistics involved in agricultural health assessments.

We believe that collecting larger and more spatially distributed datasets with inexpensive sensors is more effective than only using a few expensive sensors in very limited locations. Once sample datasets become large enough through inexpensive sensing, machine learning can be employed to calibrate ScanSpectrum spectrometers to approximate the performance of “gold standard” equipment.

Here we showcase some pictures of our devices to date. We are currently piloting our devices with our partners in several countries, and have been able to successfully demonstrate detection of particular kinds of plant stresses and illnesses. Our gun-shaped device ships in a compact, fitted carrying case complete with documentation, accessories, and spare parts. ScanSpectrum interfaces with your Android phone via Bluetooth, and reports both the spectrum and computed NDVI in real-time, while also streaming this georeferenced data to the cloud.

Software Screenshots

Below are some more pictures of the mobile app interface. Note that you must have our hardware to operate the Android app properly. So please disregard the bipolar ratings we currently have in the Play Store.

Acknowledgments, Current Status, and Contact

We are very thankful to USAID and CIMMYT Nepal for supporting our development of ScanSpectrum.

ScanSpectrum is useful. We have manufactured dozens of fully working prototypes in our makerspace lab. We have both field data and computational analyses demonstrating the ability of our low-cost spectrometer to discriminate between healthy and sick plants, and compute NDVI-based metrics at least as well as other agronomic sensors on the market that can be 20 times more expensive. We started by first studying work done at the PublicLab at CERN, and then building our own design to be more rugged and precise, with all data catalogued by custom Android and web software. We also have a variety of tips that can be attached to ScanSpectrum like gun silencers, to perform not just passive sensing but also active sensing under enclosed environmental conditions.

To move to the next level, we need scale. The need for low-cost and inexpensive sensors to fill the big data gap in agronomy is blindingly obvious to us, and we need partners that share the same vision. We seek ambitious, visionary partners who also place high priorities on data collection and innovating around ag sensing, who appreciate the value of our work, and who can build the right connections to help us take this to the next level.

If you are interested in collaborating with us, please e-mail scanspectrum@qed.ai or try our Contact Us form.