• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2005.11a
• /
• pp.317-320
• /
• 2005

본 논문에서는 퍼지 로직을 이용하여 로봇의 여러 환경변수에 따라 로봇들의 행동을 적절히 선택하는 알고리즘을 제시한다. 전략 및 전술 알고리즘으로 많이 알려진 Modular Q-학습 알고리즘은 개체의 수에 따른 상태수를 지수 함수적으로 증가시킬 뿐만 아니라, 로봇이 협력하기 위해 중재자모듈이라는 별도의 알고리즘을 필요로 한다. 그러나 앞으로 제시하는 로봇 행동의 퍼지 적정성을 고려한 로봇축구 전략 및 전술 알고리즘은 환경 변수에 따라 로봇 행동의 적절성을 퍼지 로직을 통하여 얻어내게 하였으며, 이를 이용함으로써 다수 로봇의 상호작용도 고려할 수 있게 하였다.

• Smart Structures and Systems
• /
• v.7 no.3
• /
• pp.169-184
• /
• 2011

We review the current understanding of modularity in biological motor control and its forms, and then relate this modularity to proposed modular control structures for biomimetic robots. We note the features that are different between the robotic and the biological 'designs' with features which have evolved by natural selection, and note those aspects of biology which may be counter-intuitive or unique to the biological controls as we currently understand them. Biological modularity can be divided into kinematic modularity comprised of strokes and cycles: primitives approximating a range of optimization criteria, and execution modularity comprised of kinetic motor primitives: muscle synergies recruited by premotor drives which are most often pulsatile, and which have the biomechanical effect of instantiating a visco-elastic force-field in the limb. The relations of these identified biological elements to kinematic and force-level motor primitives employed in robot control formulations are discussed.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.1136-1139
• /
• 2021

쇼핑카트 부착 모듈형 로봇 'Cart-Rider'는 어드미턴스 제어를 통한 사용자의 힘 보조 기능, 딥러닝을 활용한 네비게이션 기능, GPS 를 활용한 도난 방지 기능을 제공하는 로봇으로 대형 마트에서 발생하는 안전사고 및 쇼핑카트 도난을 예방하는 동시에 사용자에게 편의성을 제공하는 로봇이다. 또한 여러 대를 겹쳐서 보관하는 기존의 카트 시스템을 유지하고 탈부착이 용이하도록 하드웨어를 제작하여 환경에 영향을 주지 않고 유지 및 보수가 용이하도록 제작했다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.129-131
• /
• 2022

본 논문에서는 다양한 인공지능을 모듈화하고 모듈들을 서로 결합하여 서비스를 제공할 수 있는 지능형 서비스 로봇에서, 인공지능 모듈들을 라이브러리 간의 의존성을 해소하기 위한 방법 중 하나인 가상 머신의 일종인 도커(Docker)를 활용하여 컨테이너화하여 사용할 때, 인공지능 모듈 내부에서 사용하는 신경망 데이터에 해당하는 지능 모델에 대해 버전 관리를 수행하면서 클라우드 등 외부 서버를 이용하여 증강시킨 지능 모델을 공유하는 기술 개발에 대해 설명한다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2001.10b
• /
• pp.55-57
• /
• 2001

강화학습은 한 에이전트가 자신이 놓여진 환경으로부터의 보상을 최대화할 수 있는 최적의 행동 전략을 학습하는 것이다. 따라서 강화학습은 입력(상태)과 출력(행동)의 쌍으로 명확한 훈련 예들이 제공되는 교사 학습과는 다르다. 특히 Q-학습과 같은 비 모델 기반(model-free)의 강화학습은 사전에 환경에 대한 별다른 모델을 설정하거나 학습할 필요가 없으며 다양한 상태와 행동들을 충분히 자주 경험할 수만 있으면 최적의 행동전략에 도달할 수 있어 다양한 응용분야에 적용되고 있다. 하지만 실제 응용분야에서 Q-학습과 같은 강화학습이 겪는 최대의 문제는 큰 상태 공간을 갖는 문제의 경우에는 적절한 시간 내에 각 상태와 행동들에 대한 최적의 Q값에 수렴할 수 없어 효과를 거두기 어렵다는 점이다. 이런 문제점을 고려하여 본 논문에서는 로봇 축구 시뮬레이션 환경에서 각 선수 에이전트의 동적 위치 결정을 위해 효과적인 새로운 Q-학습 방법을 제안한다. 이 방법은 원래 문제의 상태공간을 몇 개의 작은 모듈들로 나누고 이들의 개별적인 Q-학습 결과를 단순히 결합하는 종래의 모듈화 Q-학습(Modular Q-Learning)을 개선하여, 보상에 끼친 각 모듈의 기여도에 따라 모듈들의 학습결과를 적응적으로 결합하는 방법이다. 이와 같은 적응적 중재에 기초한 모듈화 Q-학습법(Adaptive Mediation based Modular Q-Learning, AMMQL)은 종래의 모듈화 Q-학습법의 장점과 마찬가지로 큰 상태공간의 문제를 해결할 수 있을 뿐 아니라 보다 동적인 환경변화에 유연하게 적응하여 새로운 행동 전략을 학습할 수 있다는 장점을 추가로 가질 수 있다. 이러한 특성을 지닌 AMMQL 학습법은 로봇축구와 같이 끊임없이 실시간적으로 변화가 일어나는 다중 에이전트 환경에서 특히 높은 효과를 볼 수 있다. 본 논문에서는 AMMQL 학습방법의 개념을 소개하고, 로봇축구 에이전트의 동적 위치 결정을 위한 학습에 어떻게 이 학습방법을 적용할 수 있는지 세부 설계를 제시한다.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.10
• /
• pp.261-268
• /
• 2020

In this paper, we propose an effective software education model for youths with disability. This software education model consists of a four-step process. In the first step, it draws the education curriculum of the software education for different types of disabled youths based on the results of comparative analysis of software education field in special education curriculum. In the second step, it suggests achievement standards for effective software education for the disabled students by classifying students with intellectual disabilities and visual, hearing, and physical disabilities without any multiple disabilities. In the third step, the study developed a modular textbook comprised of unplugged activities using coding robot Albert, physical computing, and block/text coding with the reflection of the characteristic of each type of disability. In the fourth step, it applied the textbook to the school field and educated disabled students focusing on experience to allow them to think logically and by stages about different problems they face in daily lives. In addition, by analyzing the results of youths' performance evaluation and surveys, it was shown that 82.3% of developmental disabilities, 78.8% of visual impairments, 90.9% of hearing impairments, and 78.8% of physically disabilities achieved achievements above the "medium" level. These results prove that the software education model for youths with disabilities proposed in this paper is very effective in improving computational chinking of youths with disabilities.

• KIPS Transactions on Software and Data Engineering
• /
• v.8 no.12
• /
• pp.499-506
• /
• 2019

An intelligent robot should decide its behavior accordingly to the dynamic changes in the environment and user's requirements by evaluating options to choose the best one for the current situation. Many intelligent robot systems that use the Procedural Reasoning System (PRS) accomplishes such task management functions by defining the priority functions in the task model and evaluating the priority functions of the applicable tasks in the current situation. The priority functions, however, are defined locally inside of the plan, which exhibits limitation for the tasks for multi-domain services because global contexts for overall prioritization are hard to be expressed in the local priority functions. Furthermore, since the prioritization functions are not defined as an explicit module, reuse or extension of the them for general context is limited. In order to remove such limitations, we propose a policy-based meta-planning for general task management for multi-domain services, which provides the ability to explicitly define the utility of a task in the meta-planning process and thus the ability to evaluate task priorities for general context combining the modular priority functions. The ontological specification of the model also enhances the scalability of the policy model. In the experiments, adaptive behavior of a robot according to the policy model are confirmed by observing the appropriate tasks are selected in dynamic service environments.

• Journal of the Korean Society for Precision Engineering
• /
• v.30 no.6
• /
• pp.650-657
• /
• 2013

This study uses a muscle activation sensor and elbow joint model to develop an estimation algorithm for human elbow joint torque for use in a human-robot interface. A modular-type MVS (Muscle Volume Sensor) and calibration algorithm are developed to measure the muscle activation signal, which is represented through the normalization of the calibrated signal of the MVS. A Hill-type model is applied to the muscle activation signal and the kinematic model of the muscle can be used to estimate the joint torques. Experiments were performed to evaluate the performance of the proposed algorithm by isotonic contraction motion using the KIN-COM$^{(R)}$ equipment at 5, 10, and 15Nm. The algorithm and its feasibility for use as a human-robot interface are verified by comparing the joint load condition and the torque estimated by the algorithm.

• Journal of KIISE
• /
• v.42 no.3
• /
• pp.320-327
• /
• 2015

Although service robots in various fields are being commercialized, most of them have problems that depend on explicit commands by users and have difficulty to generate robust reactions of the robot in the unstable condition using insufficient sensor data. To solve these problems, we modeled mirror neuron and theory of mind systems, and applied them to a robot agent to show the usefulness. In order to implement quick and intuitive response of the mirror neuron, the proposed intention-response model utilized behavior selection networks considering external stimuli and a goal, and in order to perform reactions based on the long-term action plan of theory of mind system, we planned behaviors of the sub-goal unit using a hierarchical task network planning, and controled behavior selection network modules. Experiments with various scenarios revealed that appropriate reactions were generated according to external stimuli.

• Journal of Institute of Control, Robotics and Systems
• /
• v.19 no.2
• /
• pp.164-170
• /
• 2013

Quadrotor UAV (Unmanned Aerial Vehicle) is a flying robotic platform which has drawn lots of attention in the recent years. The attraction comes from the fact that it is able to perform agile VTOL (Vertical Take-Off Landing) and hovering functions. In addition, the efficient modular structure composed of four electric rotors makes its design easier compared to other single-rotor type helicopters. In many cases, a quadrotor often utilizes vision systems in order to obtain altitude control and navigation solution in hostile environments where GPS receivers are not working or deniable. For carrying out their successful missions, it is essential for flight control systems to have fast and stable control responses of heading angle outputs. This paper presents a Fuzzy Logic based lateral PI controller to stabilize and control the quadrotor vehicle equipped with vision systems. The advantage of using the fuzzy based PI controller lies in the fact that it could acquire a desired output response of a heading angle even in presence of disturbances and uncertainties. The performance comparison of the newly proposed Fuzzy-PI controller and the conventional PI controller was carried out with various simulation results.