A U.S. Marine places the starter into an RQ-21A Blackjack during a Weapons and Tactics Instructor course at Canon Air Defense Complex in Yuma, Ariz. The United States Marine Corps received the final of 21 Blackjack systems last June. (U.S. Marine Corps Photo)
The use of embedded and edge processing will improve the information delivery of unmanned aircraft systems (UAS) and allow them to harness the power of artificial intelligence (AI), according to a Boeing Insitu official.
Through edge processing, UAS do not have to send information to the cloud and thus may achieve greater performance, information security, and autonomy.
The comments of James McGrew, Boeing Insitu's deputy chief technology officer for technology planning and integration team lead, are featured in this month's issue of Avionics International Magazine.
Insitu builds a number of military UAS, including the ScanEagle, ScanEagle2, Scan Eagle3, the Integrator, and the latter's successor, the RQ-21A Blackjack.
"These systems are operated using interchangeable payloads and common ground support element components including launchers, Skyhook recovery systems and ground control stations," McGrew wrote. "Insitu also has a suite of Processing, Exploitation, and Dissemination (PED) tools in our Tacitview and Catalina products. As a new addition we have also extended the reach of our Integrator product line with an Extended Range SATCOM (satellite communications) kit enabling beyond line-of-site operations. It is within this family of systems that we plan to leverage the power of processing and mature AI techniques."
Insitu and other companies are embarking on efforts to embed AI on military UAS, though the timing of full and effective autonomy for such drones is uncertain.
"While it is easy to imagine a future with significant embedded AI (i.e. online learning and autonomous decision making) deployed on large swarms of UAS, there are significant challenges to adopting nondeterministic learning algorithms on unmanned systems operating in real-world situations in collaboration with our customers," according to McGrew. "As such, we are leveraging machine learning and edge processing techniques to develop tools to enhance our Family of Systems in ways that enhance operation without handing control over to 'Skkynet.'"
One example aboard the Integrator and Blackjack drones is the Hood Tech Vision AC-14 Imager payload, which uses onboard embedded processing for image stabilization and target tracking.
Embedded AI and machine learning may one day be common features for many military drones.
"AI and ML are generic terms for a wide variety of data processing, control, and optimization techniques applicable to almost any industry or system," McGrew wrote in his email. "What was considered 'AI' in the past, is considered a software application today. I’d say the possibilities are for the full spectrum from Group 1 (e.g. small fixed wing and multicopter UAS) through Group 5 (e.g. Boeing’s Air Power Teaming System), which will continue to adopt more advanced technology leading to further autonomy and allowing human operators to provide more high level input and supervisory control."
Read Frank Wolfe's full article on the topic of embedded enablement of artificial intelligence and machine learning in drones, in the December/January edition of Avionics International, publishing on Jan. 24. Click here to subscribe to our digital edition for free and get it sent to your inbox.