Files in this item

FilesDescriptionFormat

application/pdf

application/pdfRAJPUT-THESIS-2021.pdf (1MB)
(no description provided)PDF

Description

Title:Steering control and Kalman filter position estimation comparison for an autonomous underwater vehicle
Author(s):Rajput, Ayush
Advisor(s):Norris, William R
Department / Program:Industrial&Enterprise Sys Eng
Discipline:Industrial Engineering
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:M.S.
Genre:Thesis
Subject(s):Kalman Filter
Unscented Kalman Filter
Localization
Control, Fuzzy Controller
LQR Controller
Pure Pursuit Controller
Autonomous Vehicle
Underwater Vehicle
Hierarchical Rule Base Reduction
Abstract:Autonomous vehicles for sub-sea exploration are gaining in popularity. They offer longer operational time, can reach a wider and deeper area of the sea with low risk of failure. The control system and the localization system are two of the most important components that ensure the success of the mission. However, the performance of these subsystems is affected by external noise and disturbances. This thesis presents a Hierarchical rule-based reduction fuzzy controller as a solution to control systems suffering from noisy feedback and affected by external current flow disturbances. Performance comparisons with LQR and Pure Pursuit controllers show that under these conditions, the hierarchical rule-base reduction fuzzy logic controller is able to reject disturbances and sensor noises better than its counterparts. Furthermore, this research observes the performance of all three controllers under challenging path trajectories. As the complexity of the path increased, the LQR controller's performance was observed to be better than that of Fuzzy and Pure Pursuit controllers. It is suggested under uncertain dynamics and noisy sensor conditions, a fuzzy controller should be used because of its higher ability to filter out noises and reject disturbances. Challenges in localization are addressed using the Unscented version of the Kalman filter, in which reduced order dynamic model predictions are fused with measurements. When compared to the Extended Kalman filter, the Unscented Kalman Filter was observed to suppress noise much better; its performance was observed to be robust as the noise in sensor data increased. The EKF was observed to have a lower error covariance matrix value than the UKF, suggesting higher confidence in the EKF value. The UKF values were well within the acceptable limits.
Issue Date:2021-04-28
Type:Thesis
URI:http://hdl.handle.net/2142/110585
Rights Information:Copyright 2021 Ayush Rajput
Date Available in IDEALS:2021-09-17
Date Deposited:2021-05


This item appears in the following Collection(s)

Item Statistics