Aerospace Control and Guidance Systems Committee

Announcements

You must first log in to access prior meeting presentations, register for a meeting, or nominate some for the Ward Award.

If you do not have a login account, or cannot remember the email address associated with your account, please click on the Application Form link below.

Login

Site Search

Search our site:

Upcoming Events

Photos

Meeting Highlights New!

Subcommittee S

Prior Meetings

Abstracts may be viewed by anyone. Presentations are only available to active members who have logged in.

Meeting 132
(coming soon)

Home

Ward Memorial Award

Planning Advisory Board

Downloads

Constitution and By-Laws

About

History

Contact Us

← Return to agenda

Meeting	ACGS Committee Meeting 127 - San Diego, CA - November 2021
Agenda Location	9 SUBCOMMITTEE E – FLIGHT, PROPULSION, AND AUTONOMOUS VEHICLE CONTROL SYSTEMS 9.3 Safety and Learning Control
Title	Safety and Learning Control
Presenter	Aditya Gahlawat
Affiliation	UIUC
Available Downloads*	presentation
	Downloads are available to members who are logged in and either Active* or attended this meeting.
Abstract	This presentation summarizes our work that synergistically integrates data-driven learning tools with a robust adaptive control architecture. The integration aims to ensure a few fundamental properties for the persistent safety of autonomous systems. Namely: i) safety is always guaranteed by the control architecture regardless of the performance of the learning algorithm; ii) the performance and optimality depend upon the quality of learning and cannot affect the safety of the system. We derive certificates of modeling, planning, and control that can be specified in terms of hardware characteristics as CPU, GPU, actuator bandwidth, and sensor noise. The proposed architecture is modular, simplifying its analysis and system integration. The presentation further posits questions regarding notions of robustness abstractions which aim to bring closer data-driven methodologies and classical control tools. In particular, the notion of distributional robustness is examined through the lens of robust adaptive control. This line of reasoning opens up new directions in the design of control, robust to both aleatoric and epistemic uncertainties. Lifting control theoretic tools to be robust to errors in distributions themselves will open a new direction where modern data-driven methods can be equipped with the certification and analyzability that prevent their use in high-performance yet safety-aware applications.