• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2003.06a
• /
• pp.306-311
• /
• 2003

This paper presents a novel localization paradigm for mobile robots based on artificial and natural landmarks. A model-based object recognition method detects natural landmarks and conducts the global and topological localization. In addition, a metric localization method using artificial landmarks is fused to complement the deficiency of topology map and guide to action behavior. The recognition algorithm uses a modified local Zernike moments and a probabilistic voting method for the robust detection of objects in cluttered indoor environments. An artificial landmark is designed to have a three-dimensional multi-colored structure and the projection distortion of the structure encodes the distance and viewing direction of the robot. We demonstrate the feasibility of the proposed system through real world experiments using a mobile robot, KASIRI-III.

• Proceedings of the KSR Conference
• /
• 2007.05a
• /
• pp.1180-1186
• /
• 2007

This paper contains that the current emergency response procedure through the railroad accident type at domestic and foreign, and the response program which is more systematic as a result of improving weaknesses. Because the response program at home is less sufficient than developed country's, we have made an investigation into system of classification in America, British, Germany, Japan for organizing separate emergency response code. we have researched the recognition response of five senses for minimizing the accident damage in a way of spreading the emergency response type analyzed through the accident procedure according to the feature of five senses by the first recognizer. Based on the first recognitions response of five senses, the scenario has been established according to each types and the criteria of accident has been researched through classifying the recognition of five senses response level, early response level, self-response level, outside support level.

• The Transactions of the Korean Institute of Electrical Engineers C
• /
• v.50 no.7
• /
• pp.324-325
• /
• 2001

In this study, the characteristics of ultrasonic sensor fabricated with PZT-Polymer 1-3 type composites are investigated. The 3-D Underwater object recognition using the self-made ultrasonic sensor and SOFM neural network is presented. The ultrasonic sensor is satisfied with the required condition of commercial ultrasonic sensor in underwater. The 3-D underwater object recognition for the training data and the testing data are 100[100%], respectively. The experimental results have shown that the ultrasonic sensor fabricated with PZT-Polymer 1-3 type composites can be applied for sonar system.

• Journal of the Korean Society of Industry Convergence
• /
• v.18 no.1
• /
• pp.10-17
• /
• 2015

We describe a research about remote control of mobile robot based on voice command in this paper. Through real-time remote control and wireless network capabilities of an unmanned remote-control experiments and Home Security / exercise with an unmanned robot, remote control and voice recognition and voice transmission are possible to transmit on a PC using a microphone to control a robot to pinpoint of the source. Speech recognition can be controlled robot by using a remote control. In this research, speech recognition speed and direction of self-driving robot were controlled by a wireless remote control in order to verify the performance of mobile robot with two drives.

• Journal of Integrative Natural Science
• /
• v.3 no.1
• /
• pp.38-42
• /
• 2010

인간은 정보전달을 위하여 언어 이외에 동작, 표정과 같은 비언어적인 수단을 이용한다. 이러한 비언어적인 수단을 정확히 분석 할 수 있다면 인간과 컴퓨터간의 자연스럽고 지적인 인터페이스를 구축할 수 있게 된다. 본 논문은 별도의 센서를 부착하지 않은 단일 카메라 환경에서 손 형상을 입력정보로 사용하여 손 영역만을 분할한 후 자기 조직화 특징 지도(SOFM: Self Organized Feature Map) 신경망 알고리즘을 이용하여 손 형상을 인식함으로서 수화인식을 위한 보다 안정적이며 강인한 인식 시스템을 구현하고자 한다. 제안 방법으로는 피부색 정보를 이용하여 배경으로부터 손 영역만을 추출한 후 추출된 손 영역의 형상을 인식한다(전처리과정으로 모델이미지의 사이즈와 압축 및 컬러에 대한 정보를 정규화 시켰다). 또한 인식 효율을 높이기 위해 SOFM 신경망 알고리즘을 적용함으로서 보다 안정적으로 손 형상을 인식할 수 있게 되었으며, 손 형상 인식률에 대한 안전성과 정확성을 향상시킬 수 있었다. 그리고 인식된 손 형상의 의미를 텍스트로 보여줌으로서 사용자의 의사를 정확하게 전달할 수 있다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.25 no.3B
• /
• pp.565-569
• /
• 2000

We compose the two-stage recurrent neural network that returns both signals of a hidden and an output layer to the hidden layer. It is tested on the basis of syllables for Korean spoken digit from /gong/to /gu. For these experiments, we adjust the neuron number of the hidden layer, the predictive order of input data and self-recurrent coefficient of the decision state layer. By the experimental results, the recognition rate of this neural network is between 91% and 97.5% in the speaker-dependent case and between 80.75% and 92% in the speaker-independent case. In the speaker-dependent case, this network shows an equivalent recognition performance to Jordan and Elman network but in the speaker-independent case, it does improved performance.

• The Transactions of the Korea Information Processing Society
• /
• v.7 no.3
• /
• pp.971-976
• /
• 2000

In this paper a new method for head gesture recognition is proposed. A the first stage, face image data are transformed into low dimensional vectors by principal component analysis (PCA), which utilizes the high correlation between face pose images. The a self organization map(SM) is trained by the transformed face vectors, in such a that the nodes at similar locations respond to similar poses. A sequence of poses which comprises each model gesture goes through PCA and SOM, and the result is stored in the database. At the recognition stage any sequence of frames goes through the PCA and SOM, and the result is compared with the model gesture stored in the database. To improve robustness of classification, probabilistic relaxation labeling(PRL) is used, which utilizes the contextural information imbedded in the adjacent poses.

• Proceedings of the KIEE Conference
• /
• 1996.07b
• /
• pp.1237-1239
• /
• 1996

3-D object recognition and restoration independent of the translation and rotation using an ultrasonic sensor array, neural networks and invariant moment are presented. Using invariant moment vectors on the acquired $16{\times}8$ pixel data, 3-D objects can be classified by SOFM(Self Organizing Feature Map) neural networks. Invariant moment vectors kept constant independent of the translation and rotation. The experiment result shows the suggested method can be applied to the environment recognition.

• Proceedings of the KIEE Conference
• /
• 2004.05b
• /
• pp.109-112
• /
• 2004

In this study, the applicability of SOM(Self Organizing Map) algorithm to partial discharge pattern recognition have been investigated. For the purpose, using acquired data from the artificial defects in GIS, SOM algorithm which has some advantages such as data accumulation ability and the degradation trend trace ability was compared with conventionally used BP(Back Propagation) algorithm. As a result, basically BP algorithm was found out to be better than SOM algorithm. Therefore, it is needed to apply SOM algorithm in combination with BP algorithm in order to improve on-site applicability using the advantages of SOM. Also, for the pattern recognition by use of PRPDA(Phase Resolved Partial Discharge Analysis) it is required the normalization of the PRPDA graph. However, in case of the normalization both BP and SOM algorithm have shown worse results, so that it is required further study to solve the problem.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1994.06a
• /
• pp.1046-1051
• /
• 1994

It is known that SOFM has the property of effectively creating topographically the organized map of various features on input signals, SOFM can effectively be applied to the recognition of Korean phonemes. However, is isn't guaranteed that the network is sufficiently learned in SOFM algorithm. In order to solve this problem, we propose the learning algorithm combined with the conventional K-means clustering algorithm in fine-tuning stage. To evaluate the proposed algorithm, we performed speaker dependent recognition experiment using six phoneme classes. Comparing the performances of the Kohonen's algorithm with a proposed algorithm, we prove that the proposed algorithm is better than the conventional SOFM algorithm.