ACADSTAFF UGM
| Title | : | HORIZON DETECTION FOR UAV ATTITUDE BASED ON IMAGE PROCESSING APPROACH |
| Author | : |
Dr. Dyah Aruming Tyas, S.Si. (1) Ika Candradewi, S.Si., M.Cs. (2) Baskara (3) NOVELIO PUTRA INDARTO (4) HAIKAL ABDURRAHMAN (5) YOHAN ARGHA P (6) Bakhtiar Alldino Ardi Sumbodo, S.Si., M.Cs. (7) Dr. Andi Dharmawan, S.Si., M.Cs. (8) |
| Date | : | 0 2022 |
| Keyword | : | Horizon detection, Semantic segmentation, UAV Attitude, roll, pitch Horizon detection, Semantic segmentation, UAV Attitude, roll, pitch |
| Abstract | : | The UAV must have the capability for automatic stabilization to maintain its position. Automatic stabilization estimates the roll and pitch angle that makes the UAV fully autonomous. We can use AI and computer vision technology for UAV stabilization. In this study, a horizon detection system will be designed to obtain data on the orientation of the UAV relative to the earth's horizon using AI and computer vision. We proposed a method to detect the horizon line with semantic segmentation, find the horizon line candidate using dilation and bitwise operation AND, and find the horizon line by linear regression equation. Based on the evaluation of the horizon line detection test on 2 test videos, the average IoU value for the segmentation of the land area is 0.96. For the horizon line, the Root Mean Squared Error (RMSE) value for the roll is 8.50 degrees and for Pitch is 18.28%. The results of the semantic segmentation of the land area show a good performance. |
| Group of Knowledge | : | Ilmu Komputer |
| Original Language | : | English |
| Level | : | Internasional |
| Status | : |
Published
|