A Summer of Space Exploration with Intel and NASA

This summer, Intel has been collaborating with the NASA Frontier Development Lab (FDL) , an AI R&D accelerator targeting knowledge gaps useful to the space program. The NASA FDL, hosted at the SETI Institute, was established to apply AI to five specific challenges in areas relevant to the space program: Planetary Defense (defending the Earth from potentially hazardous asteroids), Space Weather (better predicting solar activity) and Space Resources (locating and accessing the resources we’ll need to go back to the moon and expand into the solar system).

Earlier this summer, we introduced you to this collaboration, and we have exciting updates to share. The NASA FDL team successfully applied the Intel® Nervana™ Deep Learning platform to automate the creation of lunar maps at our Moon’s poles – a critical step in helping both identify potential landing sites and navigation in the shadowed regions of the moon. Here, permanent darkness and extremely low temperatures make for an ideal location for water ice (and other volatiles), but highly challenging conditions for future missions that would be impossible without detailed mapping.


The Space Resource Lunar Water and Volatiles Challenge

Intel teamed up with Space Resources Luxembourg to tackle the Space Resource Mission: Lunar Water & Volatiles.  The initial purpose of the mission was to determine the location and most promising access points for vital lunar H2O, in terms of cost effectiveness and engineering constraints. 


Obstacles in Lunar Resource Exploration – and the need for AI

The Lunar Water and Volatiles team consulted engineers at the NASA Jet Propulsion Laboratory (JPL), and concluded that the main obstacle in exploring for lunar resources was the lack of accurate maps to plan missions to prospect them. The team approached the challenge by trying to solve the problem of accurately mapping lunar regions using multiple data sets. The moon’s permanently shadowed regions, like the lunar poles are of particular interest, because they have craters that are completed obscured from the Sun and have the possibility of containing lunar ice.

In an effort to create detailed lunar maps, the team examined and chose two datasets from the NASA Lunar Reconnaissance Orbiter (LRO) mission.  One data set contained optical images from the LRO Narrow Angled Camera (NAC) and the other elevation measures from the Lunar Orbiter Laser Altimeter Digital Elevation Model (LOLA DEM).

Overlaying the two datasets created highly accurate maps.  However, the registration points still needed to be aligned with the data.  The team found that the craters provided an excellent registration mark.  Ideally, craters could be identifiedfrom the DEM models, but artifacts introduced into the images from sensor data made it difficult to automate crater detection and, unfortunately, the DEM images have a lower resolution than the NAC images.

The team decided to use a Computer Vision algorithm to identify craters in both DEM and NAC data, on polar and equatorial regions of the moon.

Left: DEM, Right: NAC


Automated Lunar Crater Detection: Solving the Problem with Intel® Nervana™ Cloud

The team needed to search the image data sets for crater-like features by shape.  To an orbiting satellite, a crater may appear to be circular or ovoid (at arbitrary angle).

The team developed a computer vision algorithm based on a convolutional neural network (CNN) and analyzed images and elevation data using an adaptive convolution filter that represented the shape and shadowing of a lunar crater.

They created a training data set from of NAC and DEM image data, consisting of strips 2500m wide and 150 kilometers in length (the path traveled by an orbiting satellite). The training data consisted of square image segments containing features which might include craters.


Demo Day

The 2017 cohort concluded with a Demo Day on August 17th after eight intense weeks of problem solving.

Lunar Water and Volatiles Team used the Intel Nervana Cloud. Intel provides the technology for the researchers and data scientists to solve the world’s toughest problems.

Using the Intel Nervana Cloud, the Lunar Water and Volatiles team automated lunar crater detection with an accuracy of 98.4 percent and a speed of 1000 images per minute. By using the Intel Nervana Cloud instead of using traditional photogrammetric mapping techniques, not only did they improve accuracy but speed as well, with the CNN taking 1 minute to classify 1000 images, 100 x faster than human experts.


What’s Next?

Applying the same AI investigation techniques, we can all contribute and inspire each other to solve other space exploration problems together.  Intel is committed to continuing their partnership with NASA FDL and will be continuing to use AI to solve these mission critical problems in 2018 — and bring the excitement of space exploration to you!  The NASA FDL Automated Lunar Crater Detection project will be showcased in Intel’s booth at the NIPS conference from December 4-9 in San Diego.

Come visit and we will walk you through the process of creating your very own Automated Lunar Crater Detection program!





About Frontier Development Laboratory (FDL):

FDL is an applied artificial intelligence research accelerator and public / private partnership between NASA Ames Research Center and the SETI Institute. The program tackles knowledge gaps in space science by pairing machine learning expertise with astronomy and planetary science expertise at the PhD level.  Interdisciplinary teams address tightly defined problems and the format encourages rapid iteration and prototyping to create outputs with meaningful application to the space program.