Design, building and performance testing of GPS and computer vision combination for increasing landing precision of quad-copter drone

Abstract

This paper discusses the effort proposed for increasing accuracy of autonomous landing of quadcopter on available landing pad by combining GPS and computer vision. The main objective of this work is to recognize and to capture image from landing pad and then used it for controlling quad-copter during landing mode along with global positioning systems (GPS). The work were including design and building hardware and software of quad-copter using commercially available instruments and then tested its autonomous landing performance on predetermined landing pad. The landing pad is in the form of 3 circles black and white with similar center point on it, i.e. the inner circle (the smallest one) has radius of 25 cm, The second inner circle (the middle circle) has radius of 60 cm and the outermost circle has radius of 100 cm. The landing pad image was captured in real time and then was processed by using Canny Edge detection method involving steps, i.e. converting RGB image to grayscale, removing noise by adding blur, finding contour of image and then finally using centroid method to get center coordinate of landing pad. Testing was carried out in the outdoor environment in the sunny-day condition. Flight test was done by comparing the precision of autonomous taking off landing of quad copter using GPS only and GPS combined with computer vision algorithm on altitude 5 meters. The test results showed that landing error of quad-copter by using GPS only was 199.2 centimeters in average from the centre point compared to 42.2 centrimeters when quad copter using combination of GPS and computer vision technique.