We have designed a color mixing demonstration that showcases Mara’s capabilities to synthesize materials and follow a chemical workflow using watercolors. In this demo, we mix watercolors randomly and then collect  RGB values from images taken from a Thorlabs RGB camera and a Nano Spectrostar spectrometer to collect the wavelengths of the samples. We use two different data collection methods to confirm our camera’s ability to take accurate RGB values. For our algorithm, we are using Bayesian Optimization and Gaussian Process models to build a regression model that allows us to iteratively reach our target color.

Currently, Mara is synthesizing cobalt-based metal-organic frameworks for carbon capture. We are aiming to optimize the crystal’s morphology and size by tuning the concentrations and temperature using ML-integrated robotics.