How ncsu-yoon-lab/project-oval works
Project Type and Purpose
• Project Oval is an autonomous vehicle system aimed at operating on a university campus. It leverages a ROS2 framework running on an onboard computer (Jetson) to coordinate multiple sensors and control nodes. It not only manages vehicle motion and sensor data (GPS, camera, joystick, etc.) but also integrates external interfaces like a GUI (built with Kivy) and an AWS-hosted web server for remote control or monitoring.
• Besides the main autonomous vehicle control nodes (in the “project-oval” directory), there is an Arduino directory containing motor actuation code and a “sidewalk-segmentation” folder that appears to provide additional computer vision functionality (possibly for detecting or segmenting sidewalks for navigation).
File Structure – Architecturally
• At the top level you see folders such as “arduino”, “project-oval”, and “sidewalk-segmentation.” This separation implies a design where:
• The “project-oval” directory contains the core ROS2 Python nodes (e.g., joy, GNSS, Zed, Xbox Controller, Motor Actuation, Pure Pursuit, RTK, Server, and Camera) as well as supporting libraries.
• The “arduino” folder contains low-level controller code which the onboard computer’s Motor Actuation Node may trigger.
• “sidewalk-segmentation” contains scripts, configurations, and model files for image segmentation that might either be used as an independent module or integrated with the primary vehicle navigation system.
• Look for configuration and build files (package.xml, setup.py, requirements.txt, etc.) that underscore dependencies and environment setup typical of a ROS2 ecosystem and Python-based projects.
Final Recommendations
• Combine components wherever logical interactions occur (e.g., grouping all sensor nodes together) to avoid clutter.
• Don’t shy away from detailing even what might seem like “minor” nodes—they are critical to understanding the overall flow (especially in an autonomous system where timing and coordination are key).
• Once the draft is complete, review the intended startup sequences (as laid out in the README) to ensure every node and data path is correctly represented.