Electrical and Computer Engineering ETDs

Author

Tasha Hammond

Publication Date

7-12-2014

Abstract

Human interaction with automation is ubiquitous, occurring in many cyberphysical systems such as cell phones, automobiles, and commercial aircraft. When interacting with such systems, human users are only exposed to a simplified representation the complex system structure in the form of an interface. The human can observe system outputs and make control inputs via this interface. Problems with human-automation interaction occur when the interface does not provide enough information or provides misinformation about the underlying system, such that the human cannot determine the current state of the automation. The user's knowledge of the current system state and prediction of the next system state is required for effective operation of an automated system. In this work, formal methods are employed to analyze user-interfaces of such cyberphysical systems in order to reveal state observability problems. The cyberphysical systems are modeled as hybrid systems, for which continuous behavior emerges from the laws of physics and discrete behavior results from logical conditions and rules governing the automation. Hybrid systems with LTI continuous dynamics under collaborative control are considered, where collaborative control indicates that some events and inputs are controlled by a human operator while other events and inputs are controlled by the automation. The human user is assumed to be a special type of state observer, with additional requirements beyond a standard (automated) state observer. To reflect these additional requirements, sufficient conditions for user-observability and user-predictability of linear hybrid systems under collaborative control are developed. Algorithms are generated to evaluate a user-interface based on these conditions for user-observability and user-predictability. Then, the algorithms are applied to a hybrid system model abstraction of the longitudinal dynamics of an aircraft flight management system.

Keywords

observability, hybrid systems, control

Document Type

Thesis

Language

English

Degree Name

Electrical Engineering

Level of Degree

Masters

Department Name

Electrical and Computer Engineering

First Advisor

Oishi, Meeko

First Committee Member (Chair)

Abdallah, Chaouki

Second Committee Member

Fierro, Rafael

Share

COinS