• 대한조선학회논문집
• /
• 제56권1호
• /
• pp.58-65
• /
• 2019

The stability, reliability and efficiency of a smart ship are important issues as the interest in an autonomous ship has recently been high. An automatic collision avoidance system is an essential function of an autonomous ship. This system detects the possibility of collision and automatically takes avoidance actions in consideration of economy and safety. In order to construct an automatic collision avoidance system using reinforcement learning, in this work, the sequential decision problem of ship collision is mathematically formulated through a Markov Decision Process (MDP). A reinforcement learning environment is constructed based on the ship maneuvering equations, and then the three key components (state, action, and reward) of MDP are defined. The state uses parameters of the relationship between own-ship and target-ship, the action is the vertical distance away from the target course, and the reward is defined as a function considering safety and economics. In order to solve the sequential decision problem, the Deep Deterministic Policy Gradient (DDPG) algorithm which can express continuous action space and search an optimal action policy is utilized. The collision avoidance system is then tested assuming the $90^{\circ}$intersection encounter situation and yields a satisfactory result.

• 한국통신학회논문지
• /
• 제38A권1호
• /
• pp.1-9
• /
• 2013

본 논문에서는 CS-LBP (Center-Symmetric Local Binary Pattern) 특징과 공간 피라미드를 이용한 BoF (Bag of Features)를 생성하고 이를 랜덤 포레스트(Random Forest) 분류기에 적용하여 인간의 행동을 인식하는 알고리즘을 제안한다. BoF를 생성하기 위해 영상을 균일한 패치로 나누고, 각 패치 마다 CS-LBP 특징을 추출한다. 행동 분류 성능을 향상시키기 위해 패치들마다 추출한 특징벡터들에 대해 K-mean 클러스터링을 적용하여 코드 북을 생성한다. 본 논문에서는 영상의 지역적인 특성을 고려하기 위해 공간 피라미드 방법을 적용하고 각 공간 레벨에서 추출된 BoF에 대해 가중치를 적용하여 최종적으로 하나의 특징 벡터로 결합한다. 행동 분류를 위해 결정트리의 앙상블로 이루어진 랜덤 포레스트는 학습 단계에서 각 행동 클래스를 위한 분류 모델을 만든다. 가중 BoF가 적용된 랜덤 포레스트는 다양한 인간 행동 영상을 포함하고 있는 Standford Actions 40 데이터를 성공적으로 분류하였다. 또한 기존 방법에 비해 분류 성능이 유사하거나 우수하며, 한 장의 영상에 대해 빠른 인식속도를 보였다.

• 방송공학회논문지
• /
• 제19권2호
• /
• pp.166-173
• /
• 2014

최근 다양한 방송 및 영상 분야에서 사람의 행동을 인식하여는 연구들이 많이 이루어지고 있다. 영상은 다양한 형태를 가질 수 있기 때문에 제약된 환경에서 유용한 템플릿 방법들보다 특징점에 기반한 연구들이 실제 사용자 환경에서 더욱 관심을 받고 있다. 특징점 기반의 연구들은 영상에서 움직임이 발생하는 지점들을 찾아내어 이를 3차원 패치들로 생성한다. 이를 이용하여 영상의 움직임을 히스토그램에 기반한 descriptor(서술자)로 표현하고 학습기반의 판별기로 최종적으로 영상내에 존재하는 행동들을 인식하였다. 그러나 단일 판별기로는 다양한 행동을 인식하기에 어려움이 있다. 따라서 이러한 문제를 개선하기 위하여 최근에 다중 판별기를 활용한 연구들이 영상 판별 및 물체 검출 영역에서 사용되고 있다. 따라서 본 논문에서는 행동 인식을 위하여 support vector machine과 sparse representation을 이용한 decision-level fusion 방법을 제안하고자 한다. 제안된 논문의 방법은 영상에서 특징점 기반의 descriptor를 추출하고 이를 각각의 판별기를 통하여 판별 결과들을 획득한다. 이 후 학습단계에서 획득된 가중치를 활용하여 각 결과들을 융합하여 최종 결과를 도출하였다. 본 논문에 실험에서 제안된 방법은 기존의 융합 방법보다 높은 행동 인식 성능을 보여 주었다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제8권9호
• /
• pp.3075-3093
• /
• 2014

In this paper, we consider an advanced wireless user, called primary-secondary user (PSU) who is capable of harvesting renewable energy and connecting to both the primary network and cognitive radio networks simultaneously. Recently, energy harvesting has received a great deal of attention from the research community and is a promising approach for maintaining long lifetime of users. On the other hand, the cognitive radio function allows the wireless user to access other primary networks in an opportunistic manner as secondary users in order to receive more throughput in the current time slot. Subsequently, in the paper we propose the channel access policy for a PSU with consideration of the energy harvesting, based on a Partially Observable Markov decision process (POMDP) in which the optimal action from the action set will be selected to maximize expected long-term throughput. The simulation results show that the proposed POMDP-based channel access scheme improves the throughput of PSU, but it requires more computations to make an action decision regarding channel access.

• 한국멀티미디어학회논문지
• /
• 제13권12호
• /
• pp.1820-1831
• /
• 2010

When interacting with unknown environments, an autonomous agent needs to decide which action or action order can result in a good state and determine the transition probability based on the current state and the action taken. The traditional multiple sequential learning model requires predefined probability of the states' transition. This paper proposes a multiple sequential learning and prediction system with definition of autonomous states to enhance the automatic performance of existing AI algorithms. In sequence learning process, the sensed states are classified into several group by a set of proposed motivation filters to reduce the learning computation. In prediction process, the learning agent makes a decision based on the estimation of each state's cost to get a high payoff from the given environment. The proposed learning and prediction algorithms heightens the automatic planning of the autonomous agent for interacting with the dynamic unknown environment. This model was tested in a virtual library.

• 한국경영과학회지
• /
• 제8권1호
• /
• pp.31-40
• /
• 1983

An uncertain dynamic evolving process has been a continuing challenge to decision problems. The dynamic random variable (drv) changes which characterize such a process are very important for the decision-maker in selecting a course of action in a world that is perceived as uncertain, complex, and dynamic. Using this subjective point prediction algorithm based on a modified recursive filter, the decision-maker becomes to have periodically changing plausible points with the passage of time.

• 전자공학회논문지B
• /
• 제30B권12호
• /
• pp.51-58
• /
• 1993

This paper presents a design of the fuzzy decision maker which infers set value for fuel rate in the rotary kiln of making CaO. The fuzzy decision maker proposed are divided into two groups whose functions are different each other. The one operates when production demand is constant. The other deals with the status of varying production demand. We have chosen several variables used for composing condition and action part by investigating ingerent features of the rotary kiln and skilled operators`manual method of inferring fuel rate. Membership function of each variable was designed by analyzing experimental data and field data collected during two months. On-line operation with fuzzy rules suggested was done safely like human operators' action.

• 한국컴퓨터정보학회논문지
• /
• 제24권6호
• /
• pp.21-28
• /
• 2019

Recent years with the development of artificial intelligence and the success of the deep model, they have been deployed in all fields of computer vision. Action recognition, as an important branch of human perception and computer vision system research, has attracted more and more attention. Action recognition is a challenging task due to the special complexity of human movement, the same movement may exist between multiple individuals. The human action exists as a continuous image frame in the video, so action recognition requires more computational power than processing static images. And the simple use of the CNN network cannot achieve the desired results. Recently, the attention model has achieved good results in computer vision and natural language processing. In particular, for video action classification, after adding the attention model, it is more effective to focus on motion features and improve performance. It intuitively explains which part the model attends to when making a particular decision, which is very helpful in real applications. In this paper, we proposed a 3D dense convolutional network based on attention mechanism(ADD-Net), recognition of human motion behavior in the video.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2002년도 하계종합학술대회 논문집(3)
• /
• pp.127-130
• /
• 2002

Human's brain action can divide by recognition and intelligence. recognition is sensing voice, image and smell and Intelligence is logical judgment, inference, decision. To this concept, Define function of cerebral cortex, and apply the result. Current expert system is lack, that reasoning by cerebral cortex and thalamus, hoppocampal and so on. In this paper, With human's brain action, wish to embody human's action artificially Embody brain mechanism using Modular Neural Network, Applied this result to snake robot.

• Asia pacific journal of information systems
• /
• 제2권2호
• /
• pp.31-54
• /
• 1992

Generating an action plan and obtaining commitment to achieve it is desired in organizations, but not always achieved. Whatever the reason, there is a room for an improved approach to decision making, so that people can arrive at a common understanding of a problem and commitment to action. Those are the purposes of a Decision Conferencing(DC). A DC, one example of single workstation-based GDSSs, is a two-day session attended by a group of people who attempt to resolve important issues of concern to their organization with the help of group facilitation techniques and decision analytic computer modelling. The interchange of differing perspectives on the issues is encouraged by the facilitator who attends to group processes but does not contribute to the content of discussions. Decision analysis provides a variety of structures for modelling the differing perspectives. Information and value judgements are incorporated in these models, whose results usually reveal new, higher-level perspectives on the issues. Information technology is needed to combine the part of the model and to facilitate on-the-spot replay of results. The experimental case study in this paper shows that how a DC can help a Korea's trading company to develop new, corporate level resource allocation strategies which are based on improved consensus among competing participants.