As part of the Desert Research and Training Studies (D-RATS), JPL performs annual field tests on the ATHLETE robot in conjunction with robots from other NASA centers. While driving the robots, operators depend on high-resolution satellite images for guidance, positioning, and situational awareness. To streamline the processing of the satellite images, JPL engineers developed an application that takes advantage of the parallel nature of the workflow. JPL relies on Amazon Web Services (AWS) for this effort.
The application is built on Polyphony, which is a modular workflow orchestration framework designed to streamline the process of leveraging hundreds of nodes on Amazon Elastic Compute Cloud (Amazon EC2). By accommodating excess capacity on local machines and spare resources in the supercomputing center, Polyphony meshes perfectly with Amazon’s cloud computing. Most important, Polyphony enables the resources to work together to achieve a common goal. By using Amazon Simple Queue Service (Amazon SQS), JPL developers can deploy massive computations on Amazon EC2 by writing as little as a single class.
JPL had previously used Polyphony to validate the utility of cloud computing for processing hundreds of thousands of small images in an Amazon EC2 environment. However, JPL has adopted the cluster compute environment for processing huge images and recently processed a 3.2 giga-pixel image to support the ATHLETE robot operations in its 2010 D-RATS field test. Khawaja Shams, Senior Solution Architect, reports that “AWS’s resources completed the work in less than two hours on a cluster of 30 Cluster Compute Instances. This demonstrates a significant improvement over previous implementations.”
In addition to its support for the ATHLETE robot, Polyphony has been delivered to the Mars Science Laboratory to serve as one of the primary data processing and delivery pipelines that process data downloaded from Mars. Khawaja Shams, Senior Solution Architect, explains that the application “allowed us to process nearly 200,000 Cassini images within a few hours under $200 on AWS.” Due to the lack of elasticity available internally before switching to AWS, Khawaja explains that “we were only able to use a single machine locally and spent more than 15 days on the same task.” The efficiency and cost-savings offered by AWS has proven invaluable.