메뉴 건너뛰기
.. 내서재 .. 알림
소속 기관/학교 인증
인증하면 논문, 학술자료 등을  무료로 열람할 수 있어요.
한국대학교, 누리자동차, 시립도서관 등 나의 기관을 확인해보세요
(국내 대학 90% 이상 구독 중)
로그인 회원가입 고객센터 ENG
주제분류

추천
검색
질문

논문 기본 정보

자료유형
학술대회자료
저자정보
Chaewon Park (POSTECH(Pohang University of Science and Technology)) Soohee Han (POSTECH(Pohang University of Science and Technology))
저널정보
제어로봇시스템학회 제어로봇시스템학회 국제학술대회 논문집 ICCAS 2024
발행연도
2024.10
수록면
697 - 701 (5page)

이용수

표지
📌
연구주제
📖
연구배경
🔬
연구방법
🏆
연구결과
AI에게 요청하기
추천
검색
질문

초록· 키워드

오류제보하기
LiDAR-based place recognition holds significant potential for various applications in autonomous driving, particularly in mapping and localization. This study analyzes which loss functions are most suitable for enhancing the robustness and performance of LiDAR-based place recognition over the long term. Specifically, we train feature extraction neural networks using triplet-based loss, commonly used in LiDAR-based place recognition, and supervised contrastive loss, which has been rarely used in this context, and compare their results. Unlike traditional triplet-based learning approaches, supervised contrastive loss, which is not dependent of data sampling, excels in hard negative mining, thereby facilitating the extraction of descriminative features. Consequently, our experiments demonstrate that training feature extraction neural networks with supervised contrastive loss can improve place recognition performance compared to the widely used triplet-based loss. Experimental evaluations conducted on the NCLT dataset illustrate the performance variations according to different loss functions, particularly in terms of place recognition accuracy and robustness for long-term place recognition.

목차

Abstract
1. INTRODUCTION
2. RELATED WORK
3. METHODOLOGY
4. EXPERIMENTS
5. CONCLUSION
REFERENCES

참고문헌 (0)

참고문헌 신청

함께 읽어보면 좋을 논문

논문 유사도에 따라 DBpia 가 추천하는 논문입니다. 함께 보면 좋을 연관 논문을 확인해보세요!

이 논문의 저자 정보

최근 본 자료

전체보기

댓글(0)

0