GPSAM: Gaussian Process based smoothing and mapping using nonlinear optimization

Date

2025-1-10

Author

Keyvan, Erhan Ege

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

43
views

0
downloads

Cite This

In this thesis, we study the Simultenous Localization and Mapping problem with op- timization based routines. We rely on a object based map representation that utilizes Gaussian Processes to represent the boundaries of the objects in the environment. On top of this representation, we provide an optimization based SLAM system based on a novel smoothing scheme that is able to perform better than the existing filtering based approaches. We show that our approach has better Localization and mapping accuracy while performing better at situations where odometry information is less re- liable. When compared to previous methods, we show that our representation has a much better memory performance whilst providing stochastic information about the environment. At last, we show comparisons with previous methods to demonstrate the capabilities of our new approach.

Subject Keywords

Robotics, Mapping, Gaussian Processes, Estimation Theory

URI

https://hdl.handle.net/11511/113472

Collections

Graduate School of Natural and Applied Sciences, Thesis