Written by: Laura Dempsey
It began as an experimental pilot program for a relatively small group of researchers, only to spread across Air Force Research Laboratory (AFRL)directorates through word of mouth, successful applications, and amazing ease of use.
Bob Lee, Wright Brothers Institute (WBI) Open Innovation Project Manager, was the lead in WBI’s role as a catalyst and connecter of cloud computing usage across AFRL, bringing in experts from Google and Amazon to educate and instruct the AFRL workforce on how to use this revolutionary tool.
“We facilitated the sharing of information on best practices, how to use the cloud efficiently, and explored new things you couldn’t do behind the firewall,” Lee said. “We did that, but once the AFRL people get in, they’re pretty smart and took it and ran with it.”
The idea of cloud computing at AFRL was born of frustration. A 2018 effort to test the feasibility of using an open-source machine learning algorithm to aid wind modeling in support of precision airdrops. After expending more than 150 man hours – most of spent on seeking software approval within security regulations – the effort was stopped without testing the promising algorithm.
Later that year, WBI hosted a Sprint to explore a cloud approach. Within three days, the team got actionable results. A post-Sprint report reflects the excitement and optimism felt by the participants.
“The impact that this will have on the lab cannot be overstated in researcher time saved as well as cost savings (this test only cost $150 of cloud compute time),” the report read. “It has the potential to massively decrease research times and allow the exploration of critical new technologies such as AI and machine learning. It will also be a boon to cooperative efforts with industry and academia, enabling geographically separate partners to access the same machines and programs for quick and easy collaboration. As an organization that focuses heavily on low-TRL, unclassified basic research, this will be a game-changer for AFRL. This Sprint process allowed us to learn and test very rapidly. In this we did learn some best practices to control cost and manage access that will be shared back to AFRL/RC.”
Reading that, it is easy to see why the word started to spread. Dr. Dan Berrigan of AFRL’s Materials and Manufacturing Directorate assembled a core “test” team that went to work in the cloud, with support from industry experts brought in by WBI at no cost to WBI or the Air Force. The team quickly recognized the potential and promise of cloud computing, and Dr. Berrigan was able to clear what, to some, was an insurmountable hurdle: Leadership approval.
In March 2019, the Undersecretary of the Air Force created a “Fast Track Authorization to Operate” to speed up the software certification process. Dr. Berrigan equipped his test team with the ability to tinker, and tinker they did. Results were more than encouraging, and soon approval was granted by Mr. Michael Hanke, Director of AFRL’s Research Collaboration and Computational Directorate. Cloud computing at the Air Force Research Laboratory, where security is paramount, became a reality.
Mr. Hanke allowed experimentation on the pilot program, with numerous advantages for AFRL. They didn’t have to wait for resources, they’re able to rapidly onboard students and can give them large computer access to testing new concepts in a safe environment.
Soon, the Sensors Directorate wanted in. Computer science engineer Dr. Todd Rovito secured funding and approval to put the cloud to work with his team. Dr. Rovito saw the software certification process as antithetical to the National Defense Strategy of “building a more lethal, resilient and rapidly innovating Joint Force.”
“If successful,” Dr. Rovito explained in a memo to WBI, “it will have a huge impact on the way AFRL develops software and may bridge the valley of death in transferring out developed algorithms directly to the warfighter.”
Bob Lee organized and secured all training elements for Dr. Rovito’s team, and much like their Aerospace Systems and Materials and Manufacturing colleagues, they found cloud computing to be a very good way of doing business.
In fact, Dr. Rovito turned to G Suite, Google’s collection of cloud computing productivity and collaboration tools, for the annual enterprise-wide Scientific Advisory Board, which had been threatened with cancellation or postponement because of the COVID-19 pandemic.
Dr. Rovito gave due credit to WBI and Dr. Berrigan, saying, ““Without WBI working with Dan to pilot the use of this capability, we would not have been able to pull off the first ever virtual SAB …. Virtually impossible is now virtually possible thanks to this pioneering effort!”
Dr. Kerianne Hobbs, an aerospace engineer with Aerospace Systems, was working with the Autonomy Capability Team (ACT3). She wanted to see if that team could make use of the cloud in its highly classified and complicated collaborative projects. Dr. Hobbs sought to set up a pilot collaborative machine learning project for geographically separated collaborators in government, industry, and academia using mobile and cloud-based collaboration and computing opportunities to meet research needs.
Another WBI-organized Sprint led by Bob Lee proved that the cloud was a benefit to ACT3 in both cost savings and productivity.
“They are now able to communicate across AFRL using cloud services and Google workspace, as well as with other space researchers and autonomy researchers for exploration her area,” Lee said. “It has cost $240. The cost comparisons are amazing- This provides flex capability. If you use onsite computers, you have to buy them then you have to use them then you have to dispose of them.
“With cloud computing, you just rent it for the time you need it.”
The most recent AFRL directorate making cloud computing overtures is the 711th Human Performance Wing to see how it can enhance their research. Again, WBI is supporting, guiding and helping design the effort to meet the directorate’s specific requirements.
WBI’s Bob Lee sums it up:
“Dan Berrigan and Mr. Hanke got the original pilot done; Roz Richardson set up a pilot experiment in how to control costs and maintain accountability. Adam Gerlagh was able to try something he couldn’t put on a government computer,” Lee said.
“Now, researchers have an opportunity that costs zero. If it’s too hard or too expensive to try something, they won’t try it. Adam wanted to try this open source because he couldn’t get approval to put (his work) on the network,” he continued.
“It’s an interesting try before you buy proposition.”