Problem: From its core, solar panels use the sun as their source. Sunlight contains photons, each with a different energy and wavelength range (UV to infrared). According to the Shockley-Queisser limit, the maximum possible efficiency for solar power is 33.7%.
Solution: HelioRL uses reinforcement learning (RL) to optimize the positions and orientation of dual-axis solar panels to maximize their energy production. This hackathon solution won first place in the TKS Activate Focus Hackathon 2022.
HelioRL: Increasing Solar Panel Efficiency by Optimizing Panel Orientation
https://prod-files-secure.s3.us-west-2.amazonaws.com/01a992de-ace2-4d69-9084-925091123227/89066844-3854-4c0d-8d80-27d8f66d370c/HelioRL.pdf