The goal of this project was to design a robot that repairs lane paint while tracking lanes and detecting damaged parts of lane paint. This project won the Special Prize at the 19th Korea Intelligent Robot Contest.

Contribution: I designed a PID-like method to control the robot; participated in designing lane paint and damage detection algorithm; and designed an algorithm to paint damaged parts of lanes.

A monocular RGB camera was used to detect the lane and its damage. The input image was segmented horizontally. Then, each segment was binarized and convoluted with a step function to get the center of each lane segment. Linear regression was applied to the centers to get the line equation of the lane. The strength of the peak of the convoluted result was used for detecting the damage of lanes.