Indoor farming has had a rough go of it. Many companies have come on to the scene only to go bankrupt after a few years. There are many reasons, but the margins are razor thin and normally when the cost of electricity goes up a little, it is more than they can shoulder. An approach previously discussed was to use solar photovoltaics to control the energy costs . This particularly makes sense when you use agrivoltaics to power indoor farms . Innovation, however, can also play a large role. A new study just developed an open-source device for acquiring, correcting, and reconstructing the spectral power distribution of LED sources used in controlled environmental agriculture. Other such open source spectrometers have helped farmers outdoors.

Open Hardware for Indoor Farming Leverages a Bit of AI

Unlike direct measurement spectrometers, which are often too expensive for startup indoor farms, the new open source system employs a low-cost multispectral sensor (AS7265x, 18 channels, 410–940 nm) to acquire sparse band-integrated data, which are subsequently processed through a two-stage machine learning pipeline to infer a dense spectral power distribution representation. The sensor is integrated into an embedded platform that performs spectral acquisition, processing, wireless transmission, and remote visualization. Complementary environmental monitoring was included using a BME688 sensor to record temperature, humidity, and gas concentration, serving as a reference to detect and correlate anomalies in spectral power distribution measurements under extreme environmental conditions.

Using AI to Overcome Limitations in Open Hardware for Indoor Farming

The results of the study were interesting: Comparison with a reference spectrometer revealed non-linearities and some minor limits to the agreement between sensor data and ground-truth spectra. To address this, a correction stage based on a multilayer perceptron implemented with TensorFlow Lite Micro was developed, reducing the RMSE from 0.183 to 0.035 and improving the reliability of the data. All data were transmitted to a back-end server for processing. This is not the first time that open source AI has helped farmers - with many projects on the go like those to help farmers grow tomatoes. Spectral reconstruction was performed in the cloud using a one-dimensional convolutional neural network trained on horticultural LED spectra and physically inspired synthetic spectra representative of indoor farming as a whole. The model achieved an error of only 0.0135, confirming high precision within the target application domain and demonstrating a scalable and cost-effective solution for spectral monitoring in controlled agricultural environments. This open hardware device cuts the cost of the tool by a factor of at least 5 and can be made by anyone in the world.