- 16 views
- 23 downloads
Assessment of Road Infrastructure Design for Autonomous Vehicles using LiDAR Data
-
- Author / Creator
- Abdelkader, Maged K G
-
Despite recent advances in automotive research, a fully autonomous system operating in open, unconstrained road environments has not yet been realized. Recently, several types of autonomous vehicle (AV) failures were reported, such as run-of-the-road collisions and failure to stop for occluded road users. Such failures raised concerns about AV compatibility with existing infrastructure. As such, offloading expensive and infeasible computational workloads to smart infrastructure is an active area of research. Representatives from the infrastructure owners/operators (IOOs), the automotive industry, and academia have advocated for new ways to prepare roadways for the deployment of AVs. These recommendations aim to assess and enhance road infrastructure design for AVs by IOOs. These recommendations are threefold. First, improving vehicle-to-infrastructure connectivity (V2I), known as “talking to the road.” Second, enhancing the vehicle’s ability to monitor the driving environment, or “seeing the road.” Finally, “simplifying the road” through modifying roadway and roadside geometric design to support navigation by AVs. Following these recommendations, this thesis explores the gap between existing road infrastructure and AV technologies, specifically i) occlusion due to physical infrastructure, ii) road geometric design, iii) roadside design, iv) pavement markings, and v) traffic signs. In this dissertation, novel simulation-based and quantitative approaches were developed to segment light detection and ranging (LiDAR) data and assess road design performance for AVs using ultra-dense point clouds. The methods utilize point clouds to digitize existing roads and simulate a set of AV systems within this environment. An infrastructure surface model, efficient octree/voxel data structures, and semantic segmentation variants of deep learning methods and heuristic segmentation approaches are used to map design measures and extract locations with substandard conditions for AVs. A framework for improvement solutions is also presented. This work helps infrastructure operators and those in the AV sector make data-driven decisions regarding smart physical and digital infrastructure investments. Methods developed in this research are the first to systemically and quantitatively assess road design for AVs on a large scale.
-
- Subjects / Keywords
-
- Graduation date
- Spring 2024
-
- Type of Item
- Thesis
-
- Degree
- Doctor of Philosophy
-
- License
- This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.