Improving TRN performance over incomplete and inaccurate map regions
Abstract/Contents
- Abstract
- The primary focus of this thesis is on improving the operation of terrain relative navigation (TRN) in areas with incomplete map information. TRN is a navigation algorithm for vehicles operating in GPS denied environments. It operates by correlating measurements of the terrain to a prior terrain map in order to localize the vehicle. To address self-similarity between different sections of the terrain, TRN uses a non-parametric sample based representation of the belief distribution of vehicle locations such as a particle filter. The basic TRN measurement update requires map information for every particle in order to perform the TRN correlation step. When a subset of the particles has projected measurements that fall in an unmapped area, these off-map particles create an incomplete map problem. This thesis addresses the incomplete map problem through a new measurement update for TRN. This is called the Subcloud measurement update as it performs different updates for the on-map and off-map subsets of the particle cloud. At the core of the Subcloud measurement update is a new measurement probability for the off-map particles which is derived from the distribution of terrain heights in the off-map area. When this distribution is correct and other standard TRN assumptions are met, the Subcloud measurement update provably produces the correct posterior distribution in the on-map area. This means TRN will converge correctly when the vehicle is on-map and will correctly identify when the vehicle is off-map. For more realistic situations where the true distribution of terrain heights is not known, this thesis develops bounds on the safe operation and a modification to the Subcloud measurement update which guarantees the Subcloud measurement update will remain within these safe bounds. Finally, working with the Monterey Bay Aquarium Research Institute (MBARI), the Subcloud measurement update is demonstrated in the field for autonomous return-to-site missions with incomplete maps near the ocean floor. This thesis also presents preliminary work on improving the operation of TRN in areas where the map is inaccurate. This contribution is a pre-filter for TRN measurements that works to prevent inaccurate map information from entering the primary TRN estimator. This pre-filter exploits the presence of a multibeam sonar to run a bank of secondary TRN filters which vote to identify a consensus estimate and reject information from inaccurate map regions. This framework is tested in post processing on data from the site of an underwater volcanic eruption.
Description
Type of resource | text |
---|---|
Form | electronic resource; remote; computer; online resource |
Extent | 1 online resource. |
Place | California |
Place | [Stanford, California] |
Publisher | [Stanford University] |
Copyright date | 2022; ©2022 |
Publication date | 2022; 2022 |
Issuance | monographic |
Language | English |
Creators/Contributors
Author | Stonestrom, David Daniel |
---|---|
Degree supervisor | Gerdes, J. Christian |
Degree supervisor | Rock, Stephen M |
Thesis advisor | Gerdes, J. Christian |
Thesis advisor | Rock, Stephen M |
Thesis advisor | Schwager, Mac |
Degree committee member | Schwager, Mac |
Associated with | Stanford University, Department of Mechanical Engineering |
Subjects
Genre | Theses |
---|---|
Genre | Text |
Bibliographic information
Statement of responsibility | David Stonestrom. |
---|---|
Note | Submitted to the Department of Mechanical Engineering. |
Thesis | Thesis Ph.D. Stanford University 2022. |
Location | https://purl.stanford.edu/yd145xd1376 |
Access conditions
- Copyright
- © 2022 by David Daniel Stonestrom
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...