James Parr, co-director of the NASA Frontier Development Lab (FDL) shares how NVIDIA GPUs and deep learning can help detect, characterize and deflect asteroids.
The FDL hosted 12 standout graduate students for an internship to take on the White House’s Asteroid Grand Challenge, an ongoing program that aims to get researchers to “find all asteroid threats to human populations and know what to do about them.”
“When an near-earth object is coming toward earth, it’s really useful to figure out its shape. At the moment, we use radar to do that and we get a very low-res radar image,” mentions Parr. “We used machine learning to figure out how to turn the two-dimensional, very sparse data into something that is actually three-dimensional so we can figure out the shape of the object in an extremely quick amount of time. Doing this by hand would take weeks and we managed to do it in a few hours.”
To help find meteorites in the field, the students designed an autonomous drone, equipped with a Jetson TX1 to enhance the machine vision capabilities of the device. Then using TITAN X GPUs, the researchers developed deep learning models to build an automated meteorite detection system. Read more about the challenge on the NVIDIA blog.
Share your GPU-accelerated work with us at http://nvda.ws/2cpa2d4.
Watch more scientists and researchers share how accelerated computing is benefiting their work at http://nvda.ws/2dbscA7
Developer Spotlight: Defending the Planet Against Asteroids with Artificial Intelligence
Mar 15, 2017
Discuss (0)
Related resources
- GTC session: Cybersecurity Developer Day
- GTC session: Colossal-AI Platform: Bridging Gaps in AI Model Training and Deployment
- GTC session: Overview of Jetson Software and Developer Tools
- NGC Containers: MATLAB
- SDK: DRIVE Constellation
- Webinar: Accelerate AV Development with DGX Cloud and NVIDIA AI Enterprise