CAMBRIDGE, MA—One of the biggest challenges in developing self-driving cars and other modes of autonomous mobility is creating onboard artificial intelligence (AI) software. Given the complexity of the effort, the software must enable a vehicle to navigate the environment and detect and avoid people, buildings and other obstacles.
AI wasn’t born in a vacuum, however. Decades before self-driving cars became common, a similar technology called autonomous systems was already at work guiding spacecraft to safe landings, drones to drop-off points and undersea vehicles to detect mine finds. Tasks common to autonomous systems such as visual perception, navigation, decision-making and pattern recognition are just as important to AI.
Now two companies with overlapping interests in AI and autonomous systems are marking a milestone in their five-year relationship. Draper has a decades-long heritage in autonomous systems, having provided the capability to commercial customers and government programs. The engineering innovation company licensed vision-aided navigation components of its autonomous technology to Shield AI in 2016.
Shield AI understands autonomous vehicles from a unique perspective. The defense-technology company uses software to enable collaborative teams of unmanned aerial systems (UAS) and aircraft to operate in GPS- and RF-denied environments. Customers of the company’s technology include businesses, the military and R&D organizations.
Frank Serna, principal director of strategic systems at Draper, says vision-aided navigation is critical to making self-driving mobility safe, secure, affordable, sustainable and accessible for everyone. “Draper occupies a special position within the technology ecosystem that gives the company deep expertise in what’s required—and how to deliver—technologies that will perform in autonomous systems,” Serna said.
Serna added: “It is no easy feat to make a startup succeed. It is even harder to do so in the autonomous vehicle industry, which is unique in terms of scope, required resources and demanded skillset, but also possibilities. We are gratified that Draper’s technology is being applied to autonomous vehicles by Shield AI.”
Draper’s advanced vision-aided navigation, designed initially for unmanned aerial vehicles, does not rely on external infrastructure, such as GPS, detailed maps of the environment or motion capture systems. Draper’s vision-aided navigation uses various sensing and algorithm configurations and hardware to give a vehicle agile maneuvering and improved reliability and safety.
Andrew Reiter, a Shield AI co-founder, worked on one of Draper’s most effective robotic guidance systems earlier in his career when he was a Draper employee. Reiter said: “When you’re building a new autonomous system, there are so many components to develop, and GPS-denied navigation is one of the biggest. Draper let us hit the ground running, overcome the navigation challenge and focus on the autonomy that delivered value for our customers.”