최근 GAN(Generative Adversarial Network) 등장 이후 얼굴 표정 재연(face reenactment)의 연구가 활발해지고 있다. 얼굴 표정 재연은 입력으로 주어진 얼굴 이미지를 원하는 표정의 이미지 혹은 표정 정보를 갖는 벡터(vector)을 입력으로 주어 원하는 표정으로 합성하는 기술이다. 본 논문은 GAN 아키텍쳐(architecture)를 기반으로 회전 모듈 (rotate module)과 다양한 각도의 게임 캐릭터 표정을 표정 정보를 갖는 AUs(Action Units) vector를 통해 재연시키 는 방법을 제안한다. 입력으로 다양한 각도의 게임 캐릭터 얼굴이 주어지면 회전 모듈을 통해 정면화(frontalization) 시킨 이미지를 합성한다. 이를 통해, 다양한 각도의 게임 캐릭터들은 각도의 영향에서 벗어날 수 있다. 정면화 이미지는 원하는 표정으로 합성하기 위해 표정 정보를 갖는 AU벡터와 함께 생성자(generator)에 입력으로 주어진다. 이 때, 표정 정보를 갖고 있는 벡터는 AUs를 사용함으로써 다양한 표정과 세기(intensity)를 표현할 수 있다. 생성자는 표정 정보에 대한 관심 지역을 의미하는 관심 마스크(attention mask)를 생성하고 색상 정보를 의미하는 색상 마스크(color mask)를 생성한다. 이를 통해, 게임 캐릭터의 특징과 기타 부착물을 보존하며 표정을 재연한 이미지를 합 성할 수 있다. 관심 마스크와 색상 마스크를 이용하여 원하는 표정으로 재연한 재연 이미지를 합성하고 다시 회전 모듈을 통해 기존의 입력 이미지의 각도로 재회전하여 원하는 결과 이미지를 얻을 수 있다.

Walking method based zero moment position algorithms that can guarantee the stability of the biped walking robot while walking, but it moves the legs for the stability of the walking in a way that is not related to energy conservation. Walking method using ZMP can cause low battery efficiency and load on leg joints. The walking method using the passive walking, which is a natural and efficient method of walking, can reduce the load on the joints of the robot by using the method without using the inertia that occurs when walking and reduced control elements and efficient use of battery. In this paper, a biped robot with an actuator based on the principle of passive dynamic walker mechanism is applied to a passive walking algorithm. In order to solve the problem of stabilization of the posture during walking, the posture was stabilized by using the swing motion of the arm. and the walking movement of the robot was observed using the AHRS sensor applied to the robot .It was confirmed that the posture can be stabilized based on measured values using AHRS.

Recently, smart factories have attracted much attention as a result of the 4th Industrial Revolution. Existing factory automation technologies are generally designed for simple repetition without using vision sensors. Even small object assemblies are still dependent on manual work. To satisfy the needs for replacing the existing system with new technology such as bin picking and visual servoing, precision and real-time application should be core. Therefore in our work we focused on the core elements by using deep learning algorithm to detect and classify the target object for real-time and analyzing the object features. We chose YOLO CNN which is capable of real-time working and combining the two tasks as mentioned above though there are lots of good deep learning algorithms such as Mask R-CNN and Fast R-CNN. Then through the line and inside features extracted from target object, we can obtain final outline and estimate object posture.

This paper presents a vision-based relative pose estimation algorithm and its validation through both numerical and hardware experiments. The algorithm and the hardware system were simultaneously designed considering actual experimental conditions. Two estimation techniques were utilized to estimate relative pose; one was a nonlinear least square method for initial estimation, and the other was an extended Kalman Filter for subsequent on-line estimation. A measurement model of the vision sensor and equations of motion including nonlinear perturbations were utilized in the estimation process. Numerical simulations were performed and analyzed for both the autonomous docking and formation flying scenarios. A configuration of LED-based beacons was designed to avoid measurement singularity, and its structural information was implemented in the estimation algorithm. The proposed algorithm was verified again in the experimental environment by using the Autonomous Spacecraft Test Environment for Rendezvous In proXimity (ASTERIX) facility. Additionally, a laser distance meter was added to the estimation algorithm to improve the relative position estimation accuracy. Throughout this study, the performance required for autonomous docking could be presented by confirming the change in estimation accuracy with respect to the level of measurement error. In addition, hardware experiments confirmed the effectiveness of the suggested algorithm and its applicability to actual tasks in the real world.

Side scanning sonar (SSS) provides valuable information for robot navigation. However using the side scanning sonar images in the navigation was not fully studied. In this paper, we use range data, and side scanning sonar images from UnderWater Simulator (UWSim) and propose measurement models in a feature based simultaneous localization and mapping (SLAM) framework. The range data is obtained by echosounder and sidescanning sonar images from side scan sonar module for UWSim. For the feature, we used the A-KAZE feature for the SSS image matching and adjusting the relative robot pose by SSS bundle adjustment (BA) with Ceres solver. We use BA for the loop closure constraint of pose-graph SLAM. We used the Incremental Smoothing and Mapping (iSAM) to optimize the graph. The optimized trajectory was compared against the dead reckoning (DR).

One of the most frequently performed tasks in human-robot interaction (HRI), intelligent vehicles, and security systems is face related applications such as face recognition, facial expression recognition, driver state monitoring, and gaze estimation. In these applications, accurate head pose estimation is an important issue. However, conventional methods have been lacking in accuracy, robustness or processing speed in practical use. In this paper, we propose a novel method for estimating head pose with a monocular camera. The proposed algorithm is based on a deep neural network for multi-task learning using a small grayscale image. This network jointly detects multi-view faces and estimates head pose in hard environmental conditions such as illumination change and large pose change. The proposed framework quantitatively and qualitatively outperforms the state-of-the-art method with an average head pose mean error of less than 4.5° in real-time.

We present a region-based approach for accurate pose estimation of small mechanical components. Our algorithm consists of two key phases: Multi-view object co-segmentation and pose estimation. In the first phase, we explain an automatic method to extract binary masks of a target object captured from multiple viewpoints. For initialization, we assume the target object is bounded by the convex volume of interest defined by a few user inputs. The co-segmented target object shares the same geometric representation in space, and has distinctive color models from those of the backgrounds. In the second phase, we retrieve a 3D model instance with correct upright orientation, and estimate a relative pose of the object observed from images. Our energy function, combining region and boundary terms for the proposed measures, maximizes the overlapping regions and boundaries between the multi-view co-segmentations and projected masks of the reference model. Based on high-quality co-segmentations consistent across all different viewpoints, our final results are accurate model indices and pose parameters of the extracted object. We demonstrate the effectiveness of the proposed method using various examples.

In this study, a model-referenced underwater navigation algorithm is proposed for high-precise underwater navigation using monocular vision near underwater structures. The main idea of this navigation algorithm is that a 3D model-based pose estimation is combined with the inertial navigation using an extended Kalman filter (EKF). The spatial information obtained from the navigation algorithm is utilized for enabling the underwater robot to navigate near underwater structures whose geometric models are known a priori. For investigating the performance of the proposed approach the model-referenced navigation algorithm was applied to an underwater robot and a set of experiments was carried out in a water tank.

In this paper, we propose a method for estimating the pose of the camera using a rectangle feature utilized for the visual SLAM. A warped rectangle feature as a quadrilateral in the image by the perspective transformation is reconstructed by the Coupled Line Camera algorithm. In order to fully reconstruct a rectangle in the real world coordinate, the distance between the features and the camera is needed. The distance in the real world coordinate can be measured by using a stereo camera. Using properties of the line camera, the physical size of the rectangle feature can be induced from the distance. The correspondence between the quadrilateral in the image and the rectangle in the real world coordinate can restore the relative pose between the camera and the feature through obtaining the homography. In order to evaluate the performance, we analyzed the result of proposed method with its reference pose in Gazebo robot simulator.

The versatility of a human hand is what the researchers eager to mimic. As one of the attempt, the redundant degree of freedom in the human hand is considered. However, in the force domain the redundant joint causes a control issue. To solve this problem, the force control method for a redundant robotic hand which is similar to the human is proposed. First, the redundancy of the human hand is analyzed. Then, to resolve the redundancy in force domain, the artificial minimum energy point is specified and the restoring force is used to control the configuration of the finger other than the force in a null space. Finally, the method is verified experimentally with a commercial robot hand, called Allegro Hand with a force/torque sensor.

This paper proposes a novel method for detection of hand raising poses from images acquired from a single camera attached to a mobile robot that navigates unknown dynamic environments. Due to unconstrained illumination, a high level of variance in human appearances and unpredictable backgrounds, detecting hand raising gestures from an image acquired from a camera attached to a mobile robot is very challenging. The proposed method first detects faces to determine the region of interest (ROI), and in this ROI, we detect hands by using a HOG-based hand detector. By using the color distribution of the face region, we evaluate each candidate in the detected hand region. To deal with cases of failure in face detection, we also use a HOG-based hand raising pose detector. Unlike other hand raising pose detector systems, we evaluate our algorithm with images acquired from the camera and images obtained from the Internet that contain unknown backgrounds and unconstrained illumination. The level of variance in hand raising poses in these images is very high. Our experiment results show that the proposed method robustly detects hand raising poses in complex backgrounds and unknown lighting conditions.

This paper proposes a pose-graph based SLAM method using an upward-looking camera and artificial landmarks for AGVs in factory environments. The proposed method provides a way to acquire the camera extrinsic matrix and improves the accuracy of feature observation using a low-costcamera. SLAM is conducted by optimizing AGV’s explored path using the artificial landmarks installed on the ceiling at various locations. As the AGV explores, the pose nodes are added based on the certain distance from odometry and the landmark nodes are registered when AGV recognizes the fiducial marks. As a result of the proposed scheme, a graph network is created and optimized through a G2O optimization tool so that the accumulated error due to the slip is minimized. The experiment shows that the proposed method is robust for SLAM in real factory environments.

This paper presents a method of improving the pose recognition accuracy of objects by using Kinect sensor. First, by using the SURF algorithm, which is one of the most widely used local features point algorithms, we modify inner parameters of the algorithm for efficient object recognition. The proposed method is adjusting the distance between the box filter, modifying Hessian matrix, and eliminating improper key points. In the second, the object orientation is estimated based on the homography. Finally the novel approach of Auto-scaling method is proposed to improve accuracy of object pose estimation. The proposed algorithm is experimentally tested with objects in the plane and its effectiveness is validated.

본 논문에서는 키보드나 마우스를 이용하지 않고 손 포즈나 동작으로 직관적인 사용자 인터 페이스를 제공하기 위한 실시간 손 포즈 인식 방법을 제안한다. 먼저 깊이 카메라 입력영상에서 왼손과 오른손의 영역을 분할 및 잡음 보정 후 각 손 영역에 대하여 손 회전각과 손 중심점을 계산한다. 그리고 손 중심점에서 일정간격으로 원을 확장해 나가면서 손 경계 교차점의 중간 지점을 구해 손가락 관절점과 끝점을 검출한다. 마지막으로 앞서 구한 손 정보와 이전 프레임의 손 모델간의 매칭을 수행하여 손 포즈를 인식한 후 다음 프레임을 위하여 손 모델을 갱신한다. 본 방법은 연속된 프레임간의 시간 일관성을 이용하여 이전 프레임의 손 모델 정보를 통하여 은닉된 손가락의 예측이 가능하다. 양손을 사용하여 은닉된 손가락을 가진 다양한 손 포즈에 대해 실험한 결과 제안 방법은 평균 95% 이상의 정확도로 32 fps 이상의 성능을 보였다. 제안 방법은 프리젠테이션, 광고, 교육, 게임 등의 응용분야에서 비접촉식 입력 인터페이스로 사용될 수 있다.

A mobile manipulator is a system with a robotic manipulator mounted on top of a mobile base. It has both indoor and outdoor applications for transporting or transferring materials. When a user gives commands, they are usually at high levels such as “move the object to the table,” or “tidy the room.” By intelligently decomposing these complex commands into several subtasks, the mobile manipulator can perform the tasks with a greater efficiency. One of the crucial subtasks for these commands is the pick‐and‐place task. For the mobile manipulator, selection of a good base position and orientation is essential to accomplishing this task. This paper presents an algorithm that determines one of the position and orientation of a mobile manipulator in order to complete the pickand‐ place task without human intervention. Its effectiveness are shown for a mobile manipulator with 9 degrees‐of‐freedom in simulations

In this paper, we present a global localization and position error compensation method in a known indoor environment using magnet hall sensors. In previous our researches, it was possible to compensate the pose errors of xe, ye, θe correctly on the surface of indoor environment with magnets sets by regularly arrange the magnets sets of identical pattern. To improve the proposed method, new strategy that can realize the global localization by changing arrangement of magnet pole is presented in this paper. Total six patterns of the magnets set form the unique landmarks. Therefore, the virtual map can be built by using the six landmarks randomly. The robots search a pattern of magnets set by rotating, and obtain the current global pose information by comparing the measured neighboring patterns with the map information that is saved in advance. We provide experimental results to show the effectiveness of the proposed method for a differential drive wheeled mobile robot.