eYSIP-2017_Evaluating_Blackline_Robot

This project develops a video analytic software system for automatic evaluation of videos of robots following black line in the e-yantra competitions.The videos are taken using unknown camera while the robot moves over predefined arena of fixed dimensions. The arena as well as the robot are labelled with ArUco markers. Robot motion trajectory was extracted from the video by doing camera calibration, pose estimation and localization. Next, the trajectory is evaluated by computing the average deviation from an ideal path. The trajectory extraction was implemented in C++ using OpenCV 3.1.0 while trajectory evaluation was done in MATLAB.Experiments were carried out to measure the accuracy which showed that the robot can be localized to about 1 cm accuracy on a 6X6 feet arena from a video captured on mobile phones. An automatic evaluation of videos of black line following robots was carried out and found to match the human evaluation.