Awards
METU Graduation Project Competition (3rd Place)
METU Computer Engineering Department ∙ June 2024
  • Led a 5-person team in developing RoboDetection, a robot dog simulation project designed to aid rescue teams in disaster zones. The system, controlled via a remote web interface, autonomously detects and tracks individuals while generating a real-time 2D map of the environment.
  • Established the ROS connection between the server and the simulation environment using Husarnet VPN, enabling remote movement commands to be sent to the robot dog via a web interface.
  • Subscribed to the camera topic in the simulation to retrieve camera data and prepared it for transmission to the web interface for real-time monitoring.
  • Implemented the 2D map generation feature using the gmapping package in ROS, allowing the robot dog to create a real-time map of its environment.
TEKNOFEST Artificial Intelligence in Transportation Competition (2nd Place)
TEKNOFEST ∙ September 2021
  • In this competition, our task was to detect various objects and areas from UAV footage, including vehicles (automobiles, trucks, motorbikes, buses, etc.), pedestrians, as well as specialized zones like Flying Car Parking (FCP) areas and Flying Ambulance Landing (FAL) areas.
  • Tested the SuperGlue model for detecting Flying Car Parking (FCP) and Flying Ambulance Landing (FAL) areas but found that the model underperformed due to the lack of distinctive features in these areas.
  • Increased the dataset of FCP and FAL areas synthetically by placing these regions in varied backgrounds and applying color and shape augmentations using “flip” (Synthetic Image Generator).
  • Integrated the augmented FCP and FAL areas as new classes into the YOLOR-D6 model, significantly improving detection performance with satisfying results.
TUBITAK Unmanned Aerial Vehicle Competition (7th Place)
TUBITAK (The Scientific and Technological Research Council of Turkey) ∙ September 2020
  • Developed a color-based region detection algorithm for a rotary-wing UAV, enabling it to autonomously fly along a specified route, locate fire and water-intake areas through image processing, and extinguish the fire by releasing water.