| 제목 |
전기차 자율 충전을 위한 AVM 영상 기반 무선충전패드 검출 |
| 분야 |
전기ㆍ전자ㆍ통신 |
| 언어 |
Korean |
| 저자 |
봉은정(충북대학교), 이동규(충북대학교), 기석철(충북대학교) |
| Key Words |
AVM(어라운드 뷰 모니터링), Calibration(캘리브레이션), Semantic segmentation(의미론적 분할), Object tracking(객체 추적), Kalman filter(칼만 필터) |
| 초록 |
This paper proposes a method of recognizing a wireless charging platform environment based on an AVM image to determine a charging situation and detect the center coordinates of a wireless charging pad. With this technology, it is expected that true unmanned autonomous driving will be possible when a system by which electric vehicle-based, self-driving cars can repeat self-charging and driving is completed. An AVM image was created through calibration, and a learning dataset was generated based on the AVM image acquired from the wireless charging platform. A recognition algorithm for the wireless charging platform environment was implemented using the HRNet model, and a linear Kalman filter-based object tracking and a relationship-based object position coordinate estimation algorithm were implemented. The performance of each algorithm was evaluated using its own dataset, 3D PointCloud Map, and localization techniques. The performance evaluation of the relationship-based object estimation algorithm yielded an average error of 0.07 m. |
| 원문(PDF) |
다운로드
Journal Site
|
