• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.667-670
• /
• 2020

최근에는 증강현실, 로봇공학 등의 분야에서 객체의 위치 검출 이외에도, 객체의 자세에 대한 추정이 요구되고 있다. 객체의 자세 정보가 포함된 데이터셋은 위치 정보만 포함된 데이터셋에 비하여 상대적으로 매우 적기 때문에 인공 신경망 구조를 활용하기 어려운 측면이 있으나, 최근에 들어서는 기계학습 기반의 자세 추정 알고리즘들이 여럿 등장하고 있다. 본 논문에서는 이 가운데 Dense 6d Pose Object detector (DPOD) [11]의 구조를 기반으로 하여 가구의 조립 설명서에 그려진 가구 부품들의 자세를 추정하고자 한다. DPOD [11]는 입력으로 RGB 영상을 받으며, 해당 영상에서 자세를 추정하고자 하는 객체의 영역에 해당하는 픽셀들을 추정하고, 객체의 영역에 해당되는 각 픽셀에서 해당 객체의 3D 모델의 UV map 값을 추정한다. 이렇게 픽셀 개수만큼의 2D - 3D 대응이 생성된 이후에는, RANSAC과 PnP 알고리즘을 통해 RGB 영상에서의 객체와 객체의 3D 모델 간의 변환 관계 행렬이 구해지게 된다. 본 논문에서는 사전에 정해진 24개의 자세 후보들을 기반으로 가구 부품의 3D 모델을 2D에 투영한 RGB 영상들로 인공 신경망을 학습하였으며, 평가 시에는 실제 조립 설명서에서의 가구 부품의 자세를 추정하였다. 실험 결과 IKEA의 Stefan 의자 조립 설명서에 대하여 100%의 ADD score를 얻었으며, 추정 자세가 자세 후보군 중 정답 자세에 가장 근접한 경우를 정답으로 평가했을 때 100%의 정답률을 얻었다. 제안하는 신경망을 사용하였을 때, 가구 조립 설명서에서 가구 부품의 위치를 찾는 객체 검출기(object detection network)와, 각 개체의 종류를 구분하는 객체 리트리벌 네트워크(retrieval network)를 함께 사용하여 최종적으로 가구 부품의 자세를 추정할 수 있다.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.24 no.1
• /
• pp.47-55
• /
• 2006

This paper investigates the performance of sensor models based on satellite position and rotation angles and sensor models based on satellite orbit and attitude angles. We analyze the performance with respect to the accuracy of bundle adjustment and the accuracy of exterior orientation estimation. In particular, as one way to analyze the latter, we establish sensor models with respect to one image and apply the models to other scenes that have been acquired from the same orbit. Experiment results indicated that fer the sole purpose of bundle adjustment accuracy one could use both position-rotation models and orbit-attitude models. The accuracy of estimating exterior orientation parameters appeared similar for both models when analysis was performed based on single scene. However, when multiple scenes within the same orbital segment were used for analysis, the orbit-attitude model with attitude biases as unknowns showed the most accurate results.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.38 no.3
• /
• pp.243-248
• /
• 2010

This study addresses spacecraft moment of inertial estimation approach using Modified Rodrigues Parameters(MRP). The MRP offer advantage by avoiding singularity in Kalman Filter design for attitude determination caused by the norm constraint of quaternion parameters. Meanwhile, MRP may suffer singularity for large angular displacement, so that we designed appropriate reference attitude motion for accurate estimation. The proposed approach is expected to provide stable error covariance update with accurate spacecraft mass property estimation results.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.16 no.6
• /
• pp.1235-1241
• /
• 2012

In this paper, each driving motor for leg joints on a robot is controlled by estimating the direction of the legs measuring each joint angle and attitude angle of robot. We used quadruped working robot named TITAN-VIII in order to carry out this experimental study. 4 load cells are installed under the bottom of 4 legs to measure the pressed force on each leg while it's walking. The walking experiments of the robot were performed in 8 different conditions combined with duty factor, the length of a stride, the trajectory height of the foot and walking period of robot. The validity of attitude control for quadruped walking robot is evaluated by comparing the pressed force on a leg and the power consumption of joint driving motor. As a result, it was confirmed that the slip-condition of which the foot leave the ground late at the beginning of new period of the robot during walking process, which means the attitude control of the robot during walking process wasn't perfect only by measuring joint and attitude angle for estimating the direction of the foot.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2022.07a
• /
• pp.71-74
• /
• 2022

본 논문은 최적화알고리즘을 이용한 관절각 기반 3차원 자세 추정 기법의 정확도를 휴머노이드 로봇을 이용하여 검증하는 방법을 제안한다. 구글의 자세 추정 오픈소스 패키지인 MPP(MediaPipe Pose)로 특정자세를 취한 휴머노이드 로봇의 관절 좌표를 카메라의 픽셀 좌표로 추출한다. 추출한 픽셀 좌표를 전역최적화 방법인 uDEAS(univariate Dynamic Encoding Algorithm for Searches)를 통해 시상면과 관상면에서의 각도를 추정하고 휴머노이드 로봇의 실제 관절 각도와 비교하여 알고리즘의 정확도를 검증하는 방법을 제시한다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.05a
• /
• pp.76-77
• /
• 2022

In this paper, we propose a method to stabilize the 3D pose estimation result of a 3D volumetric data sequence by matching the pose estimation results from multi-view. Draw a circle centered on the volumetric model and project the model from the viewpoint at regular intervals. After performing Openpose 2D pose estimation on the projected 2D image, the 2D joint is matched to localize the 3D joint position. The tremor of 3D joints sequence according to the angular spacing was quantified and expressed in graphs, and the minimum conditions for stable results are suggested.

• Journal of Korean Society of Transportation
• /
• v.21 no.2
• /
• pp.155-164
• /
• 2003

현재 국내에서는 교통사고 재현 방법이 노면 흔적물에 의존하여 이루어지고 있고, 노면 흔적물이 없는 경우 교통사고 재현은 불가능하게 된다. 이러한 단점을 보완하기 위해 외국에서는 충돌모델을 이용한 교통사고 재현프로그램을 활용하고 있는 추세이다. 현재 우리나라에서 가장 많이 사용하는 교통사고 재현 프로그램 PC-CRASH를 활용하기 위해서는 각 사고마다 알맞은 충돌특성 인자 값을 사용자가 직접 입력하여 사용하여야 하나, 이를 활용할 수 있는 활용자료가 부족한 실정이다. 본 연구는 국내에서 실제 발생된 교통사고 사례에 대해 교통사고 재현 프로그램인 PC-CRASH와 충돌에 영향을 미치는 충돌 인자들로 교통사고를 재현하였고, 차량의 최종위치간 거리와 자세에 어떠한 영향을 미치는가를 알아보았다. 또한 실제 역 해석을 하기 위해 차량의 최종위치와 자세만으로 각각의 사례에 적절한 인자값을 추정할 수 있는 회귀식을 구성하였고, 통계적으로 신뢰성을 검증하였다. 사고 재현에 필요한 주요 충돌인자들의 초기치 설정시 추정식을 이용할 경우, 사고 재현 프로그램 활용 시, 시간 단축 효과를 프로그램 내부에 있는 유전자 알고리즘의 반복횟수로 추정식의 통계적 검증을 하였다.

• Journal of the Korean Society for Aviation and Aeronautics
• /
• v.19 no.4
• /
• pp.44-50
• /
• 2011

This paper presents the algorithm for attitude determination of an aircraft using binary image. An image feature vector, which is invariant to translation, scale and rotation, is constructed to capture the functional relations between the feature vector and the corresponding aircraft attitude. An iterated least squares method is suggested for estimating the attitude of given aircraft using the constructed feature vector library. Simulation results show that the proposed algorithm yields good estimates of aircraft attitude in most viewing range, although a relatively large error occurs in some limited viewing direction.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.33 no.2
• /
• pp.131-142
• /
• 2015

In order to support the development of a cost-effective river bathymetric system, this research has focused on modeling GPS observables, which are obtained by array of five single-frequency receivers (i.e., two references and three rovers) to estimate the high accurate kinematic position, and the surveying vessel altitude. Also, by applying all GPS measurements as multiple-baselines with constraining rover baselines, we derived the socalled ‘kinematic network model.’ From the model, the integer-constrained least-squares (LS) for position estimation and the implicit LS for attitude determination were implemented, while a series of simulation tests with respect to the baseline lengths around 2km performed to demonstrate its accuracy analysis. The on-the-fly (OTF) ambiguity resolution tests revealed that ninety-nine percents of time-to-fix-first ambiguity (TTFF) can be decided in less than two seconds, when the positioning accuracy of ambiguity-fixed solutions was assessed as the greater than or equal to one and two centimeters in horizontal and vertical, respectively. Comparing to the GPS-derived attitudes, the achievable accuracy gradually descended in sequence of yaw, pitch and roll due to the antenna geometric configuration. Furthermore, the RMSE values for the baseline lengths of three to six meters were within ±1′for yaw, and less than ±10′and ±20′for pitch and roll, respectively, but those of between six to fifteen meters were less than ±1′for yaw, ±5′for pitch, and ±10′for roll.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.23 no.2
• /
• pp.137-145
• /
• 2005

In this paper, we investigate the accuracy of various sensor models developed for linear pushbroom satellite images. We define the accuracy of a sensor model in two aspects: the accuracy of bundle adjustments and the accuracy of estimating exterior orientation parameters. The first accuracy has been analyzed and reported frequently whereas the second accuracy has somewhat been neglected. We argue that the second accuracy is as important as the first one. The second accuracy describes a model's ability to predict satellite orbit and attitude, which has many direct and indirect applications. Analysis was carried out on the traditional collinearity-based sensor models and orbit-based sensor models. Collinearity-based models were originally developed for aerial photos and modified for linear pushbroom-type satellite images. Orbit-based models have been used within satellite communities for satellite control and orbit determination. Models were tested with two Kompsat-1 EOC scenes and GPS-driven control points. Test results showed that orbit-based models produced better estimation of exterior orientation parameters while maintained comparable accuracy on bundle adjustments.