BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:America/New_York BEGIN:DAYLIGHT DTSTART:20260308T030000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3 TZNAME:EDT END:DAYLIGHT BEGIN:STANDARD DTSTART:20261101T010000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11 TZNAME:EST END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20261021T133715Z UID:481736EA-03DD-4A09-9329-D370B57A4BC0 DTSTART;TZID=America/New_York:20261020T190000 DTEND;TZID=America/New_York:20261020T200000 DESCRIPTION:Over the last two decades there has been significant progress i n autonomous systems developments and AI applications in aerospace. Howeve r\, these systems\, including learning components\, must have formal assur ance of their behavior before they can be safely deployed in the real-worl d. Our research provides an assurance framework for learning-enabled syste ms as a foundation for physical AI for flight. A typical architecture in t hese systems involves layers for perception\, planning and control\, where in each of these layers uses different tools and metrics for assessing rob ustness and performance. For example\, the planners -- that use vision-bas ed sensors to update the navigation and motion planning -- operate largely relying on distributionally robust stochastic optimal control\, whereas t he low-level controller can be a deterministic controller with its convent ional gain and phase (time-delay) margin. We present a new analysis framew ork for addressing this ontology challenge inherent to autonomous systems. We derive distributional robustness guarantees for deterministic L1 adapt ive controllers that can be used by any stochastic planner without facing a language barrier. The combined planner-controller framework can serve as foundation for development of certificates for V&V of learning-enabled sy stems and -- with that -- ensure a rigorous physical AI development for fl ight. An overview of different projects at our lab that build upon this fr amework will be demonstrated to show different applications. In particular \, the presentation will cover the flight tests of AirSTAR from NASA’s 2 007-2010 Aviation Safety program\, Learjet flights from Edwards AFB from 2 015-2018\, including a brief episode of Lockheed Martin’s VISTA X-62 fli ght test achieving TRL-6 in 3g flight condition\, and many others.\n\nSpea ker(s): Dr. Hovakimyan\, \n\nVirtual: https://events.vtools.ieee.org/m/575 664 LOCATION:Virtual: https://events.vtools.ieee.org/m/575664 ORGANIZER:joseph.palko@ieee.org SEQUENCE:60 SUMMARY:L1 ADAPTIVE CONTROL AND ITS TRANSITION TO PRACTICE URL;VALUE=URI:https://events.vtools.ieee.org/m/575664 X-ALT-DESC:Description: <br /><p><span lang="IT" style="font-size: 10.0pt\; font-family: 'Aptos'\,sans-serif\; mso-ascii-theme-font: minor-latin\; ms o-fareast-font-family: Aptos\; mso-fareast-theme-font: minor-latin\; mso-h ansi-theme-font: minor-latin\; mso-bidi-font-family: 'Times New Roman'\; m so-bidi-theme-font: minor-bidi\; mso-ansi-language: IT\; mso-fareast-langu age: EN-US\; mso-bidi-language: AR-SA\;">Over the last two decades there h as been significant progress in autonomous systems developments and AI app lications in aerospace.<span style="mso-spacerun: yes\;">&nbsp\; </span>Ho wever\, these systems\, including learning components\, must have formal a ssurance of their behavior before they can be safely deployed in the real- world.<span style="mso-spacerun: yes\;">&nbsp\; </span>Our research provid es an assurance framework for learning-enabled systems as a foundation for physical AI for flight.<strong> </strong>A typical architecture in these systems involves layers for perception\, planning and control\, wherein ea ch of these layers uses different tools and metrics for assessing robustne ss and performance. For example\, the planners -- that use vision-based se nsors to update the navigation and motion planning -- operate largely rely ing on distributionally robust stochastic optimal control\, whereas the lo w-level controller can be a deterministic controller with its conventional gain and phase (time-delay) margin. We present a new analysis framework f or addressing this ontology challenge inherent to autonomous systems. We d erive distributional robustness guarantees for deterministic L1 adaptive c ontrollers that can be used by any stochastic planner without facing a lan guage barrier. The combined planner-controller framework can serve as foun dation for <span style="mso-bidi-font-weight: bold\;">development of certi ficates for V&amp\;V of learning-enabled systems and -- with that -- ensur e a rigorous physical AI development for flight. An overview of different projects at our lab that build upon this framework will be demonstrated to show different applications. In particular\, the presentation will cover the flight tests of AirSTAR from NASA&rsquo\;s 2007-2010 Aviation Safety p rogram\, Learjet flights from Edwards AFB from 2015-2018\, including a bri ef episode of Lockheed Martin&rsquo\;s VISTA X-62 flight test achieving TR L-6 in 3g flight condition\, and many others.</span></span></p> END:VEVENT END:VCALENDAR