• 대한인간공학회지
• /
• 제20권3호
• /
• pp.11-25
• /
• 2001

AS human movements have the inherent property of anticipating target and can be coordinated to realize a given schedule, finger movements have stereotyped patterns during hand grasping. Finger movements have been studied in the past to find out the coordination pattern of hand joint angular movement. These studies analyzed only a few finger joints for a limited number of hand postures. This study investigated fourteen joint angles during eight hand-grasping motions to analyze the angular correlations between finger joints and to suggest motion factors which represent hand grasping. Hand grasping motions including forward arm motion were examined in ten healthy volunteers. Eight objects were used to represent real hand grasping tasks. $CyberGlove^{TM}$ and $Fasreack^{TM}$ measured hand joint angles and wrist origin. Joint angle correlations between PIJ(proximal interphalangeal joint) and MPJ(metacarpophalangeal joint) at one finger, between neighboring PIJs and MPJs were four factors related to the fast phase of hand grasping motions and eight factors related to the slow phase of hand grasping motions.

• Journal of Biosystems Engineering
• /
• 제38권2호
• /
• pp.138-148
• /
• 2013

Purpose: Determining an appropriate path is a top priority in order for a robot to maneuver in a dynamically efficient way especially in a pick-and-place task. In a non-standardized work environment, current robot arm executes its motion based on the kinematic displacements of joint variables, though resulting motion is not dynamically optimal. In this research we suggest analyzing and applying motion patterns of the human arm as an alternative to perform near optimum motion trajectory for arbitrary pick-and-place tasks. Methods: Since the motion of a human arm is very complicated and diverse, it was simplified into two links: one from the shoulder to the elbow, and the other from the elbow to the hand. Motion patterns were then divided into horizontal and vertical components and further analyzed using kinematic and dynamic methods. The kinematic analysis was performed based on the D-H parameters and the dynamic analysis was carried out to calculate various parameters such as velocity, acceleration, torque, and energy using the Newton-Euler equation of motion and Lagrange's equation. In an attempt to assess the efficacy of the analyzed human motion pattern it was compared to the virtual motion pattern created by the joint interpolation method. Results: To demonstrate the efficacy of the human arm motion mechanical and dynamical analyses were performed, followed by the comparison with the virtual robot motion path that was created by the joint interpolation method. Consequently, the human arm was observed to be in motion while the elbow was bent. In return this contributed to the increase of the manipulability and decrease of gravity and torque being exerted on the elbow. In addition, the energy required for the motion decreased. Such phenomenon was more apparent under vertical motion than horizontal motion patterns, and in shorter paths than in longer ones. Thus, one can minimize the abrasion of joints by lowering the stress applied to the bones, muscles, and joints. From the perspectives of energy and durability, the robot arm will be able to utilize its motor most effectively by adopting the motion pattern of human arm. Conclusions: By applying the motion pattern of human arm to the robot arm motion, increase in efficiency and durability is expected, which will eventually produce robots capable of moving in an energy-efficient manner.

• 한국CDE학회논문집
• /
• 제16권6호
• /
• pp.424-436
• /
• 2011

In this paper, we propose a methodology of generating a parametric segment model for human body using the Korean anthropometric data. The model is defined as an articulated body model consisted with 19 ellipsoid primitives. The primitives are joined at locations representing the physical joints of human body. A lot of previous researches have suggested methodologies of generating body models using the European or American anthropometric data, so that these models were inappropriate for engineering analyses and simulations in case of the Koreans. We defined a set of 35 body dimensions representing our segment model based on the anthropometric data of Koreans. Also we defined four key parameters of age, height, weight and waist circumference, and then we applied regression equations to associate the parameters to the aforementioned dimensions. As the results, we obtained the parametric human body segment models according to the various body types and the subject-specific models for a specific individual. The models in the various industries can be used as the base models for static and dynamic analysis considering the Koreans.

• 전기전자학회논문지
• /
• 제17권1호
• /
• pp.44-50
• /
• 2013

본 논문에서는 로봇의 팔을 기본 센서로 사용하여 별도의 센서 없이 사람을 따라다니는 햅틱 기반 로봇을 제안한다. 로봇 팔의 관절 각각에는 움직임을 생성하기 위한 모터가 있으며, 로봇 팔에 힘을 가하여 움직이면 각 관절의 각도가 변하면서 모터에서 그 각도를 출력 신호로 내보낼 수 있다. 따라서 사람이 로봇의 팔을 잡고 당기거나 밀면 로봇에서는 그 움직임을 각도로 읽어들이는 촉감 인터페이스로 활용할 수 있다. 구현된 로봇은 사람이 팔로 로봇 팔을 잡아끄는 움직임을 측정하여 움직이며 사람이 손을 잡고 이끎으로서 사람을 따라다닐 수 있다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제17권2호
• /
• pp.559-575
• /
• 2023

3D human pose estimation is widely applied in various fields, including action recognition, sports analysis, and human-computer interaction. 3D human pose estimation has achieved significant progress with the introduction of convolutional neural network (CNN). Recently, several researches have proposed the use of multiview approaches to avoid occlusions in single-view approaches. However, as the number of cameras increases, a 3D pose estimation system relying on a CNN may lack in computational resources. In addition, when a single host system uses multiple cameras, the data transition speed becomes inadequate owing to bandwidth limitations. To address this problem, we propose a distributed real-time 3D pose estimation framework based on asynchronous multiple cameras. The proposed framework comprises a central server and multiple edge devices. Each multiple-edge device estimates a 2D human pose from its view and sendsit to the central server. Subsequently, the central server synchronizes the received 2D human pose data based on the timestamps. Finally, the central server reconstructs a 3D human pose using geometrical triangulation. We demonstrate that the proposed framework increases the percentage of detected joints and successfully estimates 3D human poses in real-time.

• 대한기계학회논문집A
• /
• 제35권12호
• /
• pp.1585-1591
• /
• 2011

본 연구에서는 사용자의 손에 장착하여 손의 움직임을 측정하고 힘의 반영이 가능한 새로운 형태의 데이터 글로브(data glove)를 제안한다. 본 연구의 데이터 글로브는 인간의 외골격 구조의 분석이 기반하고 있으며 하나의 손가락 모듈은 4절기구의 조합을 통하여 1자유도로 구동이 되도록 고안되어 있다. 데이터 글로브는 펴기(extension)와 구부리기(flexion)를 할 수 있으며 내전(adduction)/외전(abduction)을 위해서 두 개의 유니버설 관절을 이용한 새로운 metacarpal joint 메커니즘을 고안하였다. 동 데이터 글로브의 유효성을 평가하기 위하여 검지손가락을 위한 구동회로와 센서를 포함한 전체 시스템을 제작하였으며 가상공간에 동적 시뮬레이션을 통해서 나타낸 물체를 조작하는 실험을 수행하였다.

• 한국산학기술학회논문지
• /
• 제20권9호
• /
• pp.525-531
• /
• 2019

인간과 컴퓨터 사이의 보다 자연스러운 상호적인 인터페이스를 효과적으로 구현하기 위해서 사람의 제스처를 활용하려는 노력이 최근 들어 지속적으로 시도되고 있다. 본 논문에서는 연속적으로 입력되는 3차원의 깊이 영상을 받아들여서 손 모델을 정의하고, 정의된 손 모델을 기반으로 사람의 손 영역을 강인하게 추출하는 알고리즘을 제시한다. 본 논문에서 제시된 알고리즘에서는 먼저 21개의 관절을 사용하여 손 모델을 정의한다. 본 논문에서 정의한 손 모델은 6개의 손바닥 관절을 포함하는 손바닥 모델과 15개의 손가락 관절을 포함하는 손가락 모델로 구성된다. 그런 다음, 입력되는 3차원의 깊이 영상을 적응적으로 이진화함으로써, 배경과 같은 비관심 영역들은 제외하고, 관심 영역인 사람의 손 영역만을 정확하게 추출한다. 실험 결과에서는 제시된 알고리즘이 연속적으로 입력되는 깊이 영상으로부터 배경과 같은 영역들은 제외하고 사람의 손 영역만을 기존의 알고리즘에 비해 약 2.4% 보다 강인하게 검출한다는 것을 보여준다. 본 논문에서 제안된 손 영역 추출 알고리즘은 제스처 인식, 가상현실 구현, 3차원 운동 게임, 수화 인식 등과 같은 컴퓨터 비전 및 영상 처리와 관련된 여러 가지의 실제적인 분야에서 유용하게 활용될 것으로 기대된다.

• 대한인간공학회지
• /
• 제13권1호
• /
• pp.3-14
• /
• 1994

In this study, the human gait trajectories during normal walking were synthesized using the inverse kinematics and optimization techniques. The synthesis is based on a gait model consisting of a torso and two legs. Each let has three segments: thigh, shank, foot, and has six degrees-of-freedom. In order to synthesize trajectories of this redundant system, the sum of angular displacements of articulating joints was selected as an objective function to be minimized. The proposed algorithm in this study is very useful for the analysis of human gait. For the gait analysis, the trajectories of four points in each leg should be measured. Hpwever, by using the algorithm, measuring the trajectories of two points is sufficient, and thus the experimental set-up can be simplified.

• 한국소음진동공학회논문집
• /
• 제20권8호
• /
• pp.761-766
• /
• 2010

The goal of this study is to analysis natural frequency for different static postures of human leg. To perform this research human leg is modeled by multi-body modeling for the musculoskeletal system. This leg model has biarticular muscles which acting on two joints and the muscles represents some of the major muscles, such as hamstring, of the upper and lower limbs. To obtain each static equilibrium position energy method is employed and to analysis natural frequency linearization method for constrained mechanical system is employed. Static equilibrium position depends on some parameter or condition such as hamstring stiffness or external force. Making a change these parameter the aim of this research can be performed.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2013년도 춘계학술대회
• /
• pp.616-617
• /
• 2013

본 논문에서는 Kinect 센서를 사용하여 인체의 특정 동작들을 감지하여 휴머노이드 로봇을 제어하는 방법을 소개한다. Kinect 센서의 depth 센서의 출력을 처리하여 인체의 각 joint 부분을 나타내는 인체 모형을 완성하였다. 인체 모형의 각 부분의 거리 및 각도를 계산하여 특정 동작을 검출하였으며 로봇에게 제어 명령을 블루투스 무선통신을 사용하여 전송한다.