Deteksi Rintangan Mengggunakan Kamera CCD untuk Aplikasi pada Traktor Tanpa Awak

Abstract

Obstacle detection is designed as a guidance for unmanned tractor to knowing the distance of obstacle in front of the tractor. The purpose of this research is to develop vision system that will be used as the guidance for unmanned tractor. CCD camera to acquire image of scene in front of tractor was placed at a tripod with height 1 meter. The position of camera is bow with 50. The image of track for tractor is captured and image was recorded in 640 x 480 pixel resolution. Six red laser pointer were added surrounding CCD camera to get obstacle image. The images were captured from 1, 2, 3, 4 and 5 meter until CCD camera cannot capture the reflection of red laser. There are some factors that effected this research: solar intensity, red laser pointer and CCD camera lens focus. The captured image were processed by image processing program for obstacle detection. The result of image processing program for obstacle detection is a prediction of an obstacle distance in front of the tractor with success rate of 100 % for 1 meter, 80 % for 2 meter, 40 % for 3 meter, 0 % for 4 meter and 5 meter, and the accuracy rate of 68 %. When the solar intensity is high, image processing of obstacle detection could not detect the obstacle from more than 3 meter.