ENPH 353 is a project course taken by UBC Engineering students centered around a machine learning competition. Students in this course practice collaboration through git on a linux-based development environment, image recognition through OpenCV, Optical Character Recognition (OCR) from scratch using Convolutional Neural Networks, robotics control through Robot Operating System and simulation on Gazebo.
The competition portion of this course consisted of programming an algorithm for a self-driving car to navigate a simulated environment in Gazebo. The car was expected to traverse on and off road, and navigate obstacles such as roundabouts, pedestrian crosswalks, and other cars. The car was tasked at reading signage along the way, which would give clues to a murder mystery. These clues were to be read using OCR off of a camera mounted at the front of the car, and uploaded to a score tracker during the competition. Each team only had one attempt to run their simulation on-stage.
My team, consisting of me and Avery Wong, placed fourth out of eighteen teams during this competition, earning the maximum possible points and only being beat out via tiebreaker for completion time.
Read about our experience through our final report, here.