http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
정밀 정찰을 목적으로 하는 4족 보행 로봇을 위한 머리모듈 진동저감 메커니즘 설계 연구
서영식(Yeongsik Seo),공경철(Kyoungchul Kong) 제어로봇시스템학회 2017 제어·로봇·시스템학회 논문지 Vol.23 No.12
Quadruped robots have been developed for various purposes including surveillance, transportation and the other service applications. However, the locomotion of a quadruped robot unavoidably accompanies translational and rotational vibrations at the robot body with multiple and varying frequency components. The vertical vibrations particularly deteriorate the surveillance accuracy of a quadruped robot. Therefore, this paper proposes a vibration suppression system of a quadruped robot for the purpose of precise surveillance. Based on the vibration analysis of a quadruped robot, called Cheetaroid-I Carrier, the proposed vibration suppression system is designed with a two-link manipulator in a bi-articular structure. The mechanical parameters of the proposed system, i.e., the frame lengths and the gravity spring constants, are optimally designed considering the vibration characteristics of the actual robot system. In the proposed system, a sky-hook controller is utilized to reject the vibration of an end-effector. The vibration suppression performance of the propose system is verified in both the time domain and the frequency domain.
LSTM을 이용한 HILS 데이터 판정 자동화 모델 개발에 대한 연구
조인정(InJeong Jo),서영식(YeongSik Seo),이형철(Heongcheol Lee) 한국자동차공학회 2020 한국자동차공학회 학술대회 및 전시회 Vol.2020 No.11
Due to the advancement of autonomous driving technology and the paradigm shift toward electric vehicle, the functions of vehicle have become diverse and complex. For effective function development and verification, car manufacturers are developing functions using the HILS system, but the complexity and diversity of functions make repeated tests inevitable. As a result, human resources are consumed in analyzing numerous test data and determining results. To overcome this problem, This paper proposed an automation model for determining HILS data using LSTM network, one of artificial networks. For model training and verification, we used 2,175 multivariate time series HILS data. Trained model classify testcase number of input data and determine Pass/Fail of input data. The performance of the proposed model was validated by performance index(accuracy, precision, recall, F1-score).
HILS 시험 결과 분석을 위한 하이브리드 신경망 개발
임완택(Wantaik Lim),오민택(Mintack Oh),서영식(Yeongsik Seo) 한국자동차공학회 2021 한국자동차공학회 부문종합 학술대회 Vol.2021 No.6
As vehicle technology rapidly changes to the form of autonomous driving and mobility and the system is complicated, simulation verification technology that combines virtual vehicle environments has been highlighted. For braking HILS system, simulation verification based on these virtual environments is possible. However, even though the evaluation is performed automatically, the resulting judgment must be verified manually by the test engineer. To solve these problems, we recently used artificial intelligence techniques that have been in the spotlight and proposed a novel hybrid deep learning model with the best validation accuracy. We analyzed the learning performance of the proposed model and demonstrated that AI judgment results are consistent by applying it to the actual HILS test results.