Cost-Effective Enablement of Automated Driving Systems on Snow-Covered Roads
Date of Award
Doctor of Philosophy
Mechanical and Aerospace Engineering
Zachary D. Asher, Ph.D.
Richard Meyer, Ph.D.
Alvis Fong, Ph.D.
Claudia Fajardo, Ph.D.
Adverse weather, automated driving, autonomous vehicles, computer vision, operational design domain, sensors
The top reason for implementing automated driving features is to increase safety. Current deployed features like lane-keeping assist and adaptive cruise control have the ability to reduce collisions by up to 50%. However, due to perception limitations, these systems are only capable of operating in strict operational design domains. For example, the occlusion of lane lines introduced during winter weather conditions causes these safety systems to disengage. Companies targeting higher automated systems often use more expensive, high-fidelity sensors that may solve problems like this, but because of the high cost and minimal business model, deployments of such scope are unsustainable. This study aims to bridge the gap between automated system capabilities in winter weather conditions while prioritizing low-cost implementations. To achieve a cost-effective proposed solution, three different research questions with various focuses will be addressed to identify (1) cost and energy consumption limitations of current deployments (2) the impact of weather on LiDAR and camera sensors, and (3) the implementation of hierarchical systems to achieve improved performance in snow-covered road conditions. The first pair of studies provide an understanding of procurement costs and operating energy consumption of automated driving systems. The second pair of studies will explore the impact of real-world weather conditions on the camera and LiDAR sensors. Lastly, the third pair of studies utilize a cost-effective sensor suite to develop a novel method of perception in snow-covered road conditions by the implementation of hierarchical systems. This collection of studies will yield a cost-effective solution for operating automated driving systems in winter weather conditions, allowing for real-world deployments with an expanded operational design domain.
Restricted to Campus until
Goberville, Nicholas Alan, "Cost-Effective Enablement of Automated Driving Systems on Snow-Covered Roads" (2022). Dissertations. 3822.