Title:
A visualization tool for perception system development and optimization
A visualization tool for perception system development and optimization
Author(s)
Warren, Michelle Dominica
Advisor(s)
Taylor, David G.
Antoniou, Antonia
Antoniou, Antonia
Editor(s)
Collections
Supplementary to
Permanent Link
Abstract
Perception systems are necessary for advanced autonomous technologies. During the integration, development, and testing phases of creating a perception system, it has been proven to be necessary to create a means for visualizing sensor data and perception information obtained from sensor fusion algorithms. The proposed research objective is to discuss the development and use of a visualization tool for the EcoCAR project, my individual contribution. To develop this visualization tool, I implemented a cluster script architecture, assisting in environment automation, that handled subscription to detected object information that may be in the form of sensor detections; sensor fused tracks; the Most Important Obstacle (MIO) track, also referred to as the lead vehicle; lane information; and publishing of markers based on the position of objects to be reflected on the visualization display with respect to the car being controlled, or the ego-vehicle. The visualization system, referred to in this thesis as Tank Viz, provided the team with a complete perception system for continued development, troubleshooting, and testing that helped lead to the team’s eventual success in Year 4 of the Mobility Challenge Connected Automated Vehicles section of the competition and provides a foundation for the team’s perception system for the 2022-2026 Electric Vehicle Challenge.
Sponsor
Date Issued
2023-05-02
Extent
Resource Type
Text
Resource Subtype
Thesis