Mid-Atlantic Section

 View Only

Hierarchical Reinforcement Learning for Air to Air Combat

  • 1.  Hierarchical Reinforcement Learning for Air to Air Combat

    Posted 30 Mar, 2021 09:37
      |   view attached
    Please join us Tuesday April 6th from 4:00-5:00pm to hear from Adrian Pope and Henry Diaz about their work on DARPA's AlphaDogfight Trials.

    Abstract:
    Artificial Intelligence ( AI) is becoming a critical component in the defense industry, as recently demonstrated by DARPA's AlphaDogfight Trials (ADT). As a participant in ADT, Lockheed Martin developed a reinforcement learning based F-16 autopilot that ultimately bested an expert human pilot. In this talk, we give an overview of the ADT program, dive into the details of our agent architecture, and also discuss how we see AI impacting the industry at large.

    Bios:
    Adrian Pope is an AI research engineer at Lockheed Martin specializing in autonomous platform control; intelligence, surveillance and reconnaissance; electronic warfare; and small unmanned aerial system integration

    Henry Diaz is an AI research engineer at Lockheed Martin with several years of experience applying cutting edge ML/RL algorithms to classification, detection and control problems. He has experience building algorithms on HPC systems and deploying them on SUAS with low SWaP C embedded HW payloads.


    ------------------------------
    Kyle Zittle
    Chair, AIAA Mid-Atlantic Section
    ------------------------------

    Attachment(s)

    pdf
    LMCO_TechTalk_April2021.pdf   400 KB 1 version