Abstract
dc:description.abstractAutonomous vehicles promise a future with fewer traffic incidents and fatalities, lower fuel consumption, and less traffic congestion. This promise hinges on the vehicle's accurate perception of the surrounding environment, which is formed by algorithms used to detect objects, track their movements, and predict their behavior in the future. However, the reliability of such algorithms is not always guaranteed, and a variety of traffic and environmental factors can significantly diminish their accuracy. To overcome this challenge and achieve the goals of vehicular autonomy, this research builds on the idea that by enabling vehicles to share and exchange information through communication networks, it becomes possible to create a collaborative perception framework that extends the range and quality of perception beyond the capabilities of individual vehicles. Specifically, this research formulates a robust, flexible, sensor- and model-agnostic framework for cooperative perception through the fusion of bird's-eye view probabilistic classification maps, introduces a new synthetic driving dataset and data generation tool that can be used to develop future perception and planning algorithms, explores how multi-agent reinforcement learning can be leveraged to enable active collaboration and cooperative perception among connected vehicles, and introduces an affordable, experimental vehicle platform for connected autonomy research that can be used to implement and test the findings of this work.
Degree
thesis:*- Name thesis:degree_name
- Doctor of Philosophy
- Level thesis:degree_level
- doctoral
- Discipline thesis:degree_discipline
- Mechanical Engineering
- Department dc:contributor.department
- Mechanical Engineering
- Grantor dc:publisher
- Virginia Tech
- Year dc:date.issued
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Mehr, Goodarz
- Chair dc:contributor.committeechair
-
- Eskandarian, Azim
- Committee members dc:contributor.committeemember
-
- Stilwell, Daniel J.
- Taheri, Saied
- Akbari Hamed, Kaveh
Subjects
dc:subject × 4Rights
dc:rights- Statement dc:rights
-
- In Copyright
- Licence dc:rights.uri
- Language dc:language.iso
- en
Identifiers
dc:identifier.*- Dc Identifier Other
- vt_gsexam:40896