Title:
TOWARDS TRACTABLE METHODS FOR FORMAL VERIFICATION OF AUTONOMY IN AEROSPACE SYSTEMS
TOWARDS TRACTABLE METHODS FOR FORMAL VERIFICATION OF AUTONOMY IN AEROSPACE SYSTEMS
Author(s)
Klett, Corbin
Advisor(s)
Feron, Eric
Chen, Yongxin
Chen, Yongxin
Editor(s)
Collections
Supplementary to
Permanent Link
Abstract
Formal verification techniques for control systems are developed and applied to realworld
aerospace systems, including experimental platforms as well as mathematical models
that contain features closely resembling those found in real systems. Though prolific
in academia, these analysis techniques are not prevalent in industry, where system-level
requirements are commonly validated by rudimentary measures of system robustness such
as gain and phase margin as well as by extensive simulation and testing. Conventional
methods have proven their efficacy for the certification of safety-critical systems but are
also incapable of exhaustively testing a system’s behaviors. Integrating more advanced
mathematical techniques into system design and analysis workflows could enable additional
autonomy capabilities, improve safety, and decrease development, operating, and
certification costs.
The verification strategies developed and demonstrated in this work rely on key results
from nonlinear systems theory, real algebraic geometry, and convex optimization. First, a
method for constructing homogeneous polynomial Lyapunov functions is presented for the
class of nonlinear systems that can be represented by a linear time-varying or a switchedlinear
system. Procedures are developed that produce improved certificates of set invariance,
bounds on peak norms, and system stability margin. Additionally, an algorithm that
uses a Lyapunov function certificate to search for a worst-case trajectory is developed and
applied to several aerospace examples, including an attitude-controlled spacecraft. Characterization
of the safe operating envelope for this spacecraft is demonstrated using Lyapunov
theory. This result is integrated into a run-time assurance algorithm, which is shown
to significantly increase the vehicle’s operational capabilities as demonstrated on an experimental
hardware platform. Finally, strategies are proposed for the formal analysis of gas
turbine engine control systems that offer advantages over some conventional practices.
Sponsor
Date Issued
2022-01-12
Extent
Resource Type
Text
Resource Subtype
Dissertation