Date of Defense
4-20-2021
Date of Graduation
5-2021
Department
Mechanical and Aerospace Engineering
First Advisor
Zachary Asher
Second Advisor
Bade Shrestha
Abstract
Autonomous vehicles consist of many complex systems, an extensive set of hardware components, and various computing algorithms that constantly work in harmony to provide selfdriving capabilities to a vehicle. This project focuses on developing a universal mount for a ZED 2 stereo camera used in data collection for autonomous driving applications. The main purpose of the mount is to reduce dashboard, windshield, and lens glare that affect the quality of data collection for autonomous driving applications. Software such as Ansys and SOLIDWORKS are used to simulate stress, strain, and thermal distributions on the mount to better understand the strong and weak points of the current mount design. Along with developing a physical product and simulations, object detection software was explored and utilized in testing the mount and glare solution. It was found that isolating the camera’s optics from the cabin of the car with the use of an enclosure called a lens hood is an effective way of eliminating dashboard glare. A detachable lens system design was found to have a small impact on reducing lens glare from light-emitting objects. The thermal simulations done on both SOLIDWORKS and Ansys have shown that the extreme weather conditions will not cause a significant impact on the mount because the material used has a high resistance to UV radiation and has a high service temperature. The static load simulations performed on both SOLIDWORKS and Ansys show that the final mount design with the selected material exhibits minimal deformation of the mount due to thermal load and the weight of the camera. For the object detection to run smoothly and classify objects in their respective categories with minimal error, it is imperative that there is almost no dashboard reflections and minimal glare coming into the lens. It is also important that the image data being recorded is clear and stable. It was shown that the final mount design addresses all the issues that affect the quality of data collection and helps improve the object detection performance significantly.
Recommended Citation
Sit, Yuan Sheng, "Autonomous Vehicle Camera Mount Application" (2021). Honors Theses. 3386.
https://scholarworks.wmich.edu/honors_theses/3386
Access Setting
Honors Thesis-Open Access
Defense Presentation
Comments
Co-authored with Parth Kadav and Noah Franklin.