This startup-like NASA project could one day bring us intelligent, self-flying drones

Last month, NASA announced that it had selected six “wild” aeronautics projects that it will test over the next two years. If the ideas are proven to be feasible—the ultimate goal—we could see significant changes in commercial airplanes. NASA said “the hope is that these creative ideas will all lead to something truly historic.”

But isn’t NASA always, so to speak, reaching for the stars? What makes these projects any more wild than a manned mission to Mars or landing a probe on a comet.

The answer is that these ideas are far, far from reality.

These are, as Transformative Aeronautics Concepts Program Doug Rohn told Fusion in a phone interview, “probably not things that are going to get into use very soon.” They’re also ideas that might not even be possible. “Some of the concepts, although they have very high benefits, they might just not prove to be feasible.” And, as project leader Ray Beach told Fusion in a phone interview, they’re wild because they’re really, really hard to pull off.

The process

Because the ideas are such a long-shot, they’re considered by NASA to be very risky. So the agency approaches allocating funds to these projects differently than to others—in essence, NASA administrators examine these projects the way Mark Cuban looks at a Shark Tank pitch.

In the press release, Rohn said “I would consider myself an early stage angel investor…the idea of the project is this is an investment process, where we’re using almost venture capital-like principles.”

But that doesn’t mean NASA is thinking like Silicon Valley investors. After all, the government’s bottom line is different from a venture capitalist’s. In Rohn’s words “Our return on investment is in knowledge and potential solutions to future challenges in aviation.”

Rohn explained that this process is in itself experimental for NASA. “We’re trying to demonstrate that an advanced idea is feasible…the way that [the teams] incubated their ideas is all kind of new.” Rohn, along with others on a panel, selected six projects out of 17.

David Krause, a structural engineer at NASA and the leader of a selected project, told Fusion in a phone interview that he thinks the process is “a very good approach to do[ing] high-risk” projects. Krause explained that, unlike with other NASA pitches, he and his team didn’t have to write a 100-page proposal. Instead, he presented his ideas in a powerpoint to a small group of people with knowledge of the subject at hand, and was able to collaborate across NASA departments.

Plus, if his project works out, the reward will be high. “If we’re successful,” he said, “there are aeronautical programs” that would pick up the project.

The aviation-disrupting wild projects

So, which wild projects made the cut? Most of them are hardware-focused, and aim to drastically change how we fly.

One project, for example, would design planes that could change shapes mid-flight. Krause wants to find better ways for planes to store energy using a hybrid electric system. That, he said, “would save a lot of fuel and cut down on carbon emissions and C02 emissions,” a long-term effort for the aviation industry. Beach is trying to build better, lighter planes that use better insulation and will make use of self-healing materials.

But perhaps the most exciting project comes from Michael Lowry, NASA’s chief scientist for Reliable Software at NASA’s Ames Research Center. In an email to Fusion, Lowry explains that he works, among other things, on projects involving artificial intelligence: “For example, how do you verify Artificial Intelligence that might be part of the extended crew on a human mission to Mars?”

For his wild NASA project, Lowry is focusing on self-piloting drones. According to him, the goals of the project are two-fold: First, he wants to build an open-source platform for Unmanned Aerial Vehicles (UAVS), called autonomy operating system (AOS), that “enables constructing intelligent UAVs out of a library of reusable apps.” The idea is to have UAV platforms, called AOS, that function the way Apple iOS and Google Android do on Smartphones. In other words, AOS would be make a drone a smart-drone.

The second goal of the project is to “determine whether it is possible for Artificial Intelligence procedures to enable a UAV to behave as a predictable certified pilot in the National Air Space—without a remote human pilot.” In plain English, Lowry wants to build a drone that will fly like a trained pilot—as he calls it, a “Pilot-in-a-Box.” Wild indeed.

Of course, these projects, including pilot-in-a-box, are still in their ideation phases. Lowry notes that “important to coordinate this with the FAA, and to work with them to determine acceptable ways of achieving software assurance.” At the moment, the FAA requires that all flying vehicles be operated by a human pilot. So we remain free of autonomous drones, for now.