베이지안 네트워크 기반 재난 대응 로봇의 탐색 목표 추론 시스템

양견모,서갑호,이종일,이석재,서진호 한국로봇학회 2018 한국재료학회지 Vol.28 No.11

In this paper, we introduce a target position reasoning system based on Bayesian network that selects destinations of robots on a map to explore compound disaster environments. Compound disaster accidents have hazardous conditions because of a low visibility and a high temperature. Before firefighters enter the environment, the robots notify information in advance, such as victim’s positions, number of victims, and status of debris of building. The problem of the previous system is that the system requires a target position to operate the robots and the firefighter need to learn how to use the robot. However, selecting the target position is not easy because of the information gap between eyewitness accounts and map coordinates. In addition, learning the technique how to use the robots needs a lot of time and money. The proposed system infers the target area using Bayesian network and selects proper x, y coordinates on the map based on image processing methods of the map. To verify the proposed system, we designed three example scenarios based on eyewetinees testimonies and compared time consumption between human and the system. In addition, we evaluate the system usability by 40 subjects.

재난환경 자율탐색을 위한 ROS기반 다중로봇 통합 관제 시스템

양견모,한종부,신훈섭,이종일,이석재,서갑호 제어·로봇·시스템학회 2019 제어·로봇·시스템학회 논문지 Vol.25 No.6

In this paper, we introduce an integrated control and operation system with multi-robot based on ROS(Robot Operating System) that selects proper robots to explore compound disaster environments. Compound disaster accidents make hazardous conditions that firefighters enter the environment immediately because of a low visibility and a high temperature. Before they start to do a rescue operation, the robots notify information in advance, such as number of victims, there positions, and status of debris of building. The problem of the previous system is that the system requires well-educated people to operate an robot. However, it is not possible or practical for the operators in every situation because learning the technique how to use the robots needs a lot of time and money. The proposed system provides useful information to do a rescue operation based on multi-robot control without operators. After a user gives target positions that they want to search, the system automatically allocates the positions to optimal robots as well as the robots start to explore disaster environment. In addition, we make the system based on ROS because we consider to extend the function of the system and to apply to various types of robots. To verify the proposed system, we implement the Gazebo simulator and the test environment in real world based on Turtlebot3. In addition, we evaluated the usability of the software interface by allowing 23 operators to test the system.

재난대응로봇 제어 안전성을 고려한 Multi-ROS기반 원격제어 시스템 개발

양견모,이재열,서갑호 제어로봇시스템학회 2021 제어로봇시스템학회 국내학술대회 논문집 Vol.2021 No.6

베이지안 네트워크 기반 재난 대응 로봇의 탐색 목표 추론시스템

양견모,서갑호,이종일,이석재,서진호 한국로봇학회 2018 로봇학회 논문지 Vol.13 No.4

In this paper, we introduce a target position reasoning system based on Bayesian network that selects destinations of robots on a map to explore compound disaster environments. Compound disaster accidents have hazardous conditions because of a low visibility and a high temperature. Before firefighters enter the environment, the robots notify information in advance, such as victim’s positions, number of victims, and status of debris of building. The problem of the previous system is that the system requires a target position to operate the robots and the firefighter need to learn how to use the robot. However, selecting the target position is not easy because of the information gap between eyewitness accounts and map coordinates. In addition, learning the technique how to use the robots needs a lot of time and money. The proposed system infers the target area using Bayesian network and selects proper x, y coordinates on the map based on image processing methods of the map. To verify the proposed system, we designed three example scenarios based on eyewetinees testimonies and compared time consumption between human and the system. In addition, we evaluate the system usability by 40 subjects.

다중 센서 기반 재난환경 통합관제 시스템을 위한 사용자 인터페이스

양견모,서갑호,함재훈,오승섭 제어로봇시스템학회 2019 제어로봇시스템학회 국내학술대회 논문집 Vol.2019 No.5

모듈형 베이지안 네트워크와 선형추론 알고리즘을 이용한 저전력 상황인식 시스템

양견모(Kyon-Mo Yang),조성배(Sung-Bae Cho) 한국정보과학회 2014 한국정보과학회 학술발표논문집 Vol.2014 No.6

도메인 지식 기반 확률모델을 이용한 스마트 TV의 상황인지 시스템

양견모(Kyon-Mo Yang),조성배(Sung-Bae Cho) 한국정보과학회 2013 정보과학회논문지 : 소프트웨어 및 응용 Vol.40 No.4

인터넷 연결과 다양한 센서가 부착된 스마트 TV는 점차 다양한 기능을 제공하고 있다. 하지만 이러한 기능을 사용하기 위해서 사용자는 각각의 기능을 학습해야 하고 상황에 맞게 선택해야 하는 문제가 있다. 상황인지 서비스는 사용자와 응용프로그램간의 상호작용을 위한 상황정보를 이용하여 사용자의 상태를 인지하고 서비스를 제공한다. 확률추론 모델인 베이지안 네트워크는 불확실한 상황에서의 상황인지를 위해 사용되었는데, 데이터가 부족할 경우에도 도메인 지식을 이용하여 네트워크를 설계할 수 있다는 장점이 있다. 본 논문에서는 스마트 TV의 상황인지 서비스를 위한 시스템을 집안환경 모델링과 상황추론으로 제공하는 시스템을 제안한다. 전자에서는 상황정보를 획득하고 서비스를 제공하며, 후자에서는 사용자, 스마트 TV, 환경으로 모듈화한 베이지안 네트워크를 사용한 상황추론 서비스를 생성한다. 제안한 방법의 추론 시간 복잡도를 LS알고리즘에서 사용한 베이지안 네트워크의 추론 시간 복잡도 계산을 통하여 계산한 결과, 기존의 베이지안 네트워크보다 감소하는 것을 확인하였다. 또한 각 서비스를 위한 대표 시나리오를 통하여 획득한 입력값으로 의도한 서비스를 수행하는 경우를 평가한 결과, 평균 89%로 계산되었다. 또한 서비스 만족도를 SUS테스트를 통하여 파악한 결과, 76.5점으로 시스템의 성능과 만족도를 입증하였다. A smart TV provides various functions using the internet connection and various sensors attached. However, there is a problem that users have to learn these functions and select them appropriately to use. A context-aware service recognizes user states and offers services using the context which is interaction information between users and applications. Bayesian network, a probabilistic model, is used for context-awareness in uncertain circumstances. Bayesian network has advantage that it can be designed with the insufficient information by using domain knowledge. In this paper, we propose a context-aware system which consists of modeling home environment and reasoning contexts. In the former, we obtain the context and offer services, and in the latter, we infer the context-awareness using modular Bayesian networks. We confirm that the proposed system has lower time complexity than the system using a monolithic network by using the LS algorithm‘s calculate method of time complexity. In addition, we assess the service performance with the input taken from the scenarios representing each service resulting in average 89% of the performance. The performance shows the accuracy of the reasoning of each networks. The satisfaction that calculated by SUS-test resulting in 76.5 points, which justifies the performance and satisfaction of the proposed system.

독립 구동형 다중 공기패드 기반 욕창 예방 매트리스 구조 연구

양견모(Kyon-Mo Yang),박민로(Min-Ro Park),구재완(Jaewan Koo),이종일(Jong-il Lee),곽동기(Donggi Gwak),이석재(Seokjae Lee),손동섭(Dong-Seop Sohn),김민규(Min-Gyu Kim),서갑호(Kap-Ho Seo) 제어로봇시스템학회 2021 제어·로봇·시스템학회 논문지 Vol.27 No.10

This study examined the structure of an independently actuated multi-air pad mattress for preventing pressure ulcers, owing to lower body pressure, in specific locations of the body where pressure sores can occur. The contributions of this study are as follows: 1) The parameters were derived and simulations were performed to design air pad units of different sizes with similar heights. 2) The mattress was manufactured by examining the bending of the folding bed, with 36 air pads and three different sizes. 3) Each air pad was independently controlled to inflate, contract, and retain to reduce the pressure on specific body positions. 4) Finally, using the fabricated prototype, the proposed designs were experimentally verified.

로봇 위치 추적을 위한 QR 마커 내부 정보 기반 카메라 위치 추정 연구

양견모(Kyon-Mo Yang),곽동기(Dong-Gi Gwak),한종부(Jong-Boo Han),함제훈(Jehun Hahm),서갑호(Kap-Ho Seo) 제어로봇시스템학회 2020 제어·로봇·시스템학회 논문지 Vol.26 No.7

Camera pose estimation based on markers may play a key role in robot localization in an indoor environment. The representative estimation method needs to correct parameters as the marker size when they want to change markers. In addition, there is a limitation of distance to detect a marker with respect to the status of the camera and the marker size. In this paper, we studied the camera pose estimation based on the information inside QR markers. To handle the problem of setting correct the marker size, the proposed system uses the QR marker that is included the information about its size. The system detects the QR marker and uses the information to estimate a camera pose adaptively. Also, we propose the nested QR marker and the hierarchical QR marker in order to increase the perceptible range of a QR marker. The nested QR marker is implemented for a drone in which the camera can move to the center of the marker. In contrast, the hierarchical QR marker is designed for a mobile robot in which the camera moves horizontally with the floor. Experimental scenario with marker, two types of markers were tried in the lab environment to test whether or not the proposed method could estimate appropriate camera pose without changing marker size by an operator. As a result, we verified the proposed method can estimate the camera pose regardless of the change of the marker size. Also, we found that the perceptible range of the propose QR markers is longer than of the single size marker.

계층적 행동선택 네트워크를 이용한 일정추천 에이전트

양견모(Kyon-Mo Yang),조성배(Sung-Bae Cho) 한국정보과학회 2012 한국정보과학회 학술발표논문집 Vol.39 No.1B

인공비서 에이전트는 일정관리 및 추천을 할 때 일정이 비어 있는 시간을 계산하는 비교적 간단한 방식을 사용하는데, 보다 유용한 추천을 위해서는 사용자의 상황과 일정의 수행 조건을 고려하여야 한다. 본 논문에서는 환경의 변화에 유연하게 대응할 수 있는 행동선택 네트워크를 사용하는 일정추천 에이전트를 개발한다. Maes가 제안한 행동선택 네트워크를 현실적인 문제에 적용하는 데는 목적과 행동 노드의 개수가 크게 늘어나면 문제가 있다. 이를 해결하기 위해 행동선택 네트워크를 모듈화 하여 목적간의 충돌을 방지하고, 모듈화를 할 때 선행 행동 연결을 통한 모듈간의 사라진 연결을 보완하며, 목적들 간의 연관관계를 표현하기 위한 상위 행동선택 네트워크를 두는 계층적 행동선택 네트워크 방식을 제안한다. 제안하는 방법을 사용하여 몇가지 시나리오에 따른 일정추천 실험을 통하여 제안한 에이전트의 유용성을 확인하였다.