Observability-Based Guidance and Sensor Placement
Year: 2014
Type: dissertation
Abstract: Control system performance is highly dependent on the quality of sensor information available.In a growing number of applications, however, the control task must be accomplished with limited sensing capabilities.This thesis addresses these types of problems from a control-theoretic point-of-view, leveraging system nonlinearities to improve sensing performance.Using measures of observability as an information quality metric, guidance trajectories and sensor distributions are designed to improve the quality of sensor information.An observability-based sensor placement algorithm is developed to compute optimal sensor configurations for a general nonlinear system.The algorithm utilizes a simulation of the nonlinear system as the source of input data, and convex optimization provides a scalable solution method.The sensor placement algorithm is applied to a study of gyroscopic sensing in insect wings.The sensor placement algorithm reveals information-rich areas on flexible insect wings, and a comparison to biological data suggests that insect wings are capable of acting as gyroscopic sensors.An observability-based guidance framework is developed for robotic navigation with limited inertial sensing.Guidance trajectories and algorithms are developed for range-only and bearing-only navigation that improve navigation accuracy.Simulations and experiments with an underwater vehicle demonstrate that the observability measure allows tuning of the navigation uncertainty.This thesis represents four years of research that was influenced by countless individuals.I would like to acknowledge the guidance of my advisor, Kristi Morgansen, who shaped my ideas while allowing me freedom in my research endeavors.My supervisory committee provided useful comments throughout the research process that sparked new ideas and refined my results.In particular, Tom Daniel was an invaluable reference, as were his students Annika Eberle and Brad Dickerson.
Author: Brian T. Hinson
Institution: University of Washington
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2619256107
