Embedded Files
Hey, I'm Andy and I've been tinkering since childhood. I love getting hands on by creating eco-friendly tech!
In the past, I developed a full-scale solar tracker that improves efficiency by orienting itself towards the sun. The solar panel's orientation is calculated using sky images and a novel hybrid algorithm! Through my research, I found that my solar tracking module improved energy efficiency by 35% while still being 33% cheaper than traditional solar trackers, drastically improving renewable energy generation.
I'm currently working on ADAPT, a low-cost, AI-based early drought warning system designed to protect small farms from flash drought. Drought affects over 40% of Earth’s land and caused over $300B in damages in 2020, with these numbers continuing to rise due to climate change. Shockingly, 79% of all small-scale farms in the U.S. lack access to drought monitoring solutions. Most existing systems are either inaccurate, retrospective, or expensive.
ADAPT addresses this challenge by utilizing physics-informed AI, where we train machine learning models on the inputs and outputs of physical climate models, retaining their physical accuracy while achieving the fast runtime of ML. Results show that this approach retains 94% of the physical model's accuracy in short-term forecasting while running over 50× faster. Because short-term forecasting is so accurate, these models are particularly effective for predicting flash drought, which can develop in under a week.
Overall, ADAPT empowers small-scale farmers with accessible drought forecasting tools, helping them prepare for water shortages and protect their crops.
Page updated
Google Sites
Report abuse