A group of talented Sixth Formers who were tasked with creating an autonomous vehicle capable of delivering packages around the school, built and programmed a robot.  Using a single camera attached to the front of the robot, it is able to retrace a route by watching a pre-recorded video of it, as demonstrated in the video.  The students also developed an Android app that allows the robot to be manually controlled using a smartphone, in order to record training videos.

The robot navigates by detecting distinctive feature points in the images it is receiving from the camera, and matching them with the features of the current frame of the training video.  This allows the median optical flow vector of the feature points to be calculated, which can be used to determine how the robot should move in order to make its view from the camera more closely match the training video.

The robot was programmed using the languages Python, C++ and Java, and uses the OpenCV computer vision library for feature matching.

Click here to watch.