Page 14 - ADT JANUARY - FEBRUARY 2023 Online Magazine

 AI/ML

© CALSPAN
BATTLEFIELDS OF THE FUTURE

IT’S ALL ABOUT HARNESSING and hierarchical agent behaviours. The single agent successfully
navigated the live plane while dynamically avoiding threats to
THE SPEED POTENTIAL AI accomplish its mission. Multi-agent RL models flew a live and virtual
AND ML OFFER Avenger to collaboratively chase a target while avoiding threats. The
hierarchical RL agent used sensor information to select courses of
 JAY MENON action based on its understanding of the world state. This demon-
strated the AI pilot’s ability to successfully process and act on live
Last December, General Atomics Aeronautical real-time information independently of a human operator to make
Systems successfully carried out multi-ob- mission-critical decisions at the speed of relevance.
jective collaborative combat missions using
artificially intelligent pilots on its Avenger GA-ASI has also announced it new Gambit Series of unmanned
unmanned aircraft system. The UAS was combat aircraft, and is specifically pitching the entire series as its
paired with “digital twin” aircraft to autono- answer to the U.S. Air Force’s future CCA requirements. Most experts
mously conduct live, virtual, and constructive agree that a mix of manned and unmanned aircraft — keeping
missions as part of GA-ASI’s efforts to advance human pilots and support operators in the loop while adding more
its Collaborative Combat Aircraft (CCA) eco- autonomy, artificial intelligence, and machine learning — is the near-
system for Autonomous Collaborative Platform term future.
(ACP) UAS using Artificial Intelligence/Machine
Learning (AI/ML). MISSION X-62

“This provides a new and innovative tool for The U.S. Air Force Research Laboratory or AFRL has invested US$15
next-generation military platforms to make million upgrading a decades-old workhorse to make it relevant
decisions under dynamic and uncertain real- for 21st century warfighter challenges. AFRL’s Autonomous Aircraft
world conditions.” The concepts demonstrated Experimentation team is using a highly modified Air Force Test Pilot
by these flights set the standard for operation- School NF-16D, an aircraft recently designated the X-62, to accel-
ally relevant mission systems capabilities on erate the development of tactical autonomy for uncrewed aircraft.
CCA platforms,” said GA-ASI Senior Director
of Advanced Programmes Michael Atwood. Matthew Niemiec, the autonomous aircraft experiment portfolio
The flight used a Reinforcement Learning (RL) lead, said the upgrades to the X-62, also known as the Variable
architecture, which demonstrated single, multi, In-flight Stability Test Aircraft, or VISTA, include software that allows
14 | JANUARY-FEBRUARY 2023 WWW.GBP.COM.SG/ADT