To validate the object detection capabilities of the vision pipeline, a decision tree was developed and tested using a red buoy as a target. The decision tree included strategically printing data that provided feedback whenever the vision pipeline successfully detected the buoy. This approach allowed the team to verify the accuracy and responsiveness of the system in real-time, ensuring that detection events were correctly identified and communicated. This iterative testing process was crucial in fine-tuning the object detection system and improving its overall reliability.

To test navigation between two buoys, a decision tree was developed and tested using two gates composed of red and green buoy towers. The decision tree was designed to calculate the midpoint between the detected buoys once a gate was seen and confirmed. Using this midpoint, the ASV was instructed to navigate to a point 3 meters beyond the midpoint, ensuring that it passes through both gates. This testing approach validated the ASV’s ability to accurately detect gates, calculate midpoints, and execute precise navigation commands for successful passage through the buoy pairs.

To test the water gun system, a decision tree and an in-house class were specifically developed for its operation. The class utilizes the Jetson’s GPIO pins to communicate with the water gun, enabling it to turn on and off as needed. Using this class, a decision tree was implemented to trigger the water gun whenever a red buoy is detected by the vision pipeline. This approach successfully validated the water gun system’s integration with the vision pipeline and ensured that it operates accurately in response to real-time object detection.