I worked as a developer on the Perception team for the final year of round 1 of the SAE AutoDrive Challenge. I improved aUToronto's 3D Object Detection range by 60% (from 25 m to 40 m) by helping develop a ground plane removal algorithm with C++ and ROS. I fine-tuned object detection parameters to improve the detection of deer, railroad crossings, and pedestrians using LiDAR clustering. Demos of the 3D Object Detection can be seen in the video below.

Additionally, I integrated Haar and SVM sign detection models for yield, no right turn, no left turn, right turn only, and left turn only signs into the C++ codebase. Due to my achievements, I was promoted to lead the Object Tracking team for round 2 of the SAE AutoDrive competition.