Applied Intuition takes flight, sets sail in acquiring EpiSci

EpiSci contributed to DARPA’s AI dogfighting algorithms, Air Combat Evolution (ACE), aboard the X-62A VISTA (shown here inverted). (Air Force photo by Kyle Brasier)

WASHINGTON — Silicon Valley startups buy each other out all the time, but rarely with implications for the Pentagon. That’s not true for the latest news from Mountain View, Calif., home of autonomy developer Applied Intuition, which is acquiring a smaller California drone firm called EpiSci. It’s a marriage of two military contractors on the cutting-edge of unmanned vehicles, meant to combine Applied’s experience in ground systems, like autonomous freight trucks and the Army’s Robotic Combat Vehicle, with EpiSci’s expertise in the sea and air, executives told Breaking Defense.

This merger is all about adding capabilities, not cutting costs, Applied co-founders Qasar Younis and Peter Ludwig said in an interview.

“We’re not letting anybody go, we’re not doing any redundancies, we’re not doing any lay-offs, or anything like that,” Younis said. “We’re keeping EpiSci intact, [because] they do the things we don’t do and we do the things they don’t do.”

“They’re actually famous within their domain,” Ludwig added. That domain gets pretty broad: EpiSci works for DARPA on unmanned air-to-air-combat, for the Navy on unmanned surface vessels and drones, for the Space Development Agency on missile-tracking satellites, and for NASA on bringing 5G connectivity to the moon, among other federal contracts.

At Applied, by contrast, “historically, all our of autonomy work has been really focused on ground, so it’s cars, trucks, it’s construction, mining,” Younis said. “We’ve done some things in the air but nothing like EpiSci. We’ve always thought, the last few years, it’d be great to get into autonomy in the air or maritime or space, [and] EpiSci actually plays in all those domains.”

To understand “the complementary nature of the companies,” Ludwig went on, it’s crucial to understand that not all robots are created equal. Teaching a computer to navigate a complex, cluttered, but essentially two-dimensional surface — like an Army RCV maneuvering around obstacles off-road — is very different from mastering movement in three dimensions with far fewer obstacles, as through the air, sea or outer space.

Roughly speaking, Ludwig told Breaking Defense, about 75 percent of the software modules required to operate autonomous vehicles can be shared across environments. These are general-purpose algorithms that ingest, interpret, and output data (such as sensor feeds) in ways that apply equally to driving, sailing and flying. But, he said, 25 percent of the code absolutely does need to be customized to the challenges of the specific “domain,” or else you get a very clever robot that can’t actually go anywhere.

Applied’s general-purpose modules are “a very mature solution for that common base layer,” he said, and it’s solid in the off-road domain. But EpiSci brings the “domain-specific algorithms … that actually work in that three-dimensional space,” Ludwig said.

The two companies are different in other ways as well. Applied, though several years younger than EpiSci, has grown faster and bigger (its latest valuation is $6 billion), has sponsored DC-area conferences, made highly public pitches to the Pentagon, and it’s delivered commercial vehicles and software in use by “thousands of engineers globally,” Younis said. By contrast, he continued, “EpiSci’s historically been a research organization, and that’s where their real strengths are.”

The trick for Applied will be to bring their ambition, scale, and ground-system expertise to EpiSci’s deep technical knowledge of its domains, without diluting either partner’s core competencies in the process.