Abstract
dc:description.abstractThe demand for autonomous vehicles (AVs) is rising across both military and civilian sectors. These unmanned systems offer numerous advantages, such as improved efficiency, safety, and adaptability. Addressing this demand requires the development of resilient and versatile autonomous vehicles crucial for the transport and reconnaissance markets. The sensory perception of autonomous vehicles of any kind is paramount to their ability to navigate and localize in their environment. Factors such as sensor noise, erroneous readings, and deliberate attacks should all be considered when developing a robust autonomous system. This work aims to quantify the degradation of sensor data which causes mapping algorithms to fail and properly localize. In this thesis, we explore five different simulated LIDAR perturbation models and their effects on mapping indoor and outdoor locations. The noise models are categorized into two types: \emph{fake} and \emph{real} point returns. A similarity metric is utilized to quantify the degradation of the resulting point clouds. An advantage of this approach, over implementing perturbations in physical environments, is the ability to test challenging or impractical perturbations on a simulated system. Our findings confirm that increased levels of noise correlate with elevated errors in mapping. We discuss the process of cascading failures and the additional overlaid topography that is produced. We also discovered that certain types of sensor noise affect indoor mapping more than outdoor, particularly when the noise is localized. In future research, we plan to investigate methods to physically implement the noise models employed in this study and to develop strategies for mitigating their impact on autonomous navigation.
Degree
thesis:*- Name thesis:degree_name
- Master of Science in Electrical and Computer Engineering (MS)
- Level thesis:degree_level
- Open Access Master's Thesis
- Discipline thesis:degree_discipline
- Department of Electrical and Computer Engineering
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Kangas, Dylan J.
- Committee members dc:contributor.committeemember
-
- Anthony Pinar
- Evan Lucas
- Contributors dc:contributor
-
- Timothy Havens
- Jin Choi
Subjects
dc:subject × 8Identifiers
dc:identifier.*- Identifier
- https://digitalcommons.mtu.edu/etdr/1761