Auto - Pilot : Autonomous Control of a remote controlled helicopter

An auto-pilot system was created which could track and control a Syma S107 RC mini infrared helicopter in real-time. The helicopter was tracked using two LEDs attached to it, one situated on the front of the helicopter and one situated on the tail. Image processing and object tracking algorithms were used to track the position of the helicopter. In order to perform real-time tracking, the methods used had to be very efficient. Thresholding was used for segmentation, so that the co-ordinates of the two LEDs on the helicopter could be extracted. CamShift was used to track the helicopter and use its position to limit the search space for the LEDs. The orientation of the helicopter was found using a combination of the distance between the two LEDs and the distance of the helicopter from the camera. The Kinect’s depth sensor was used to determine the distance of the helicopter. The helicopter was controlled by reverse engineering its infrared protocol. An infrared LED was attached to an Arduino Uno board in order to send the infrared signals and control the helicopter from the computer. A closed-loop feedback system was created where the helicopter’s position and orientation was used to make it perform certain movements, including hovering and flying in a horizontal square. ACM Computing Classification System Classification Thesis classification under the ACM Computing Classification System (1998 version, valid through 2012) I.2.9 [Artificial Intelligence] : Robotics—autonomous vehicles I.4.8 [Image Processing and Computer Vision] : Scene Analysis—tracking, colour General Terms: Algorithms, Performance