• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.12
• /
• pp.2730-2736
• /
• 2010

Recently, various networks have been introduced in the car of the internal and external sides. These have been integrated by one HMI(Human Machine Interface) to control devices of each network and provide information service. The existing vehicle navigation system, providing GPS based vehicle positioning service, has been included to these integrated networks as a default option. The GPS has been used to the most universal device to provide position information by using satellites' signal. But It is impossible to provide the position information when the GPS can't receive the satellites' signal in the area of tunnel, urban canyon, or forest canopy. Thus, this paper propose and implement the method of measuring vehicle position by using the sensing data of internal CAN network and external Wi-Fi network of the integrated car navigation circumstances when the GPS doesn't work normally. The results obtained by implementation shows the proposed method works well by map matching.

• Journal of Institute of Control, Robotics and Systems
• /
• v.12 no.9
• /
• pp.926-934
• /
• 2006

This paper proposes intelligent control of a virtual walking machine that can generate infinite floor for various surfaces and can provide proprioceptive feedback of walking to a user. This machine allows users to participate in a life-like walking experience in virtual environments with various terrains. The controller of the machine is implemented hierarchically, at low-level for robust actuator control, at mid-level fur platform control to compensate the external forces by foot contact, and at high-level control for generating walking trajectory. The high level controller is suggested to generate continuous walking on an infinite floor for various terrains. For the high level control, each independent platform follows a man foot during the swing phase, while the other platform moves back during single stance phase. During double limb support, two platforms manipulate neutral positions to compensate the offset errors generated by velocity changes. This control can, therefore, satisfy natural walking conditions in any direction. Transition phase between the swing and the stance phases is detected by using simple switch sensor system, while human foot motions are sensed by careful calibration with a magnetic motion tracker attached to the shoe. Experimental results of walking simulations at level ground, slope, and stairs, show that with the proposed machine, a general person can walk naturally on various terrains with safety and without any considerable disturbances. This interface can be applied to various areas such as VR navigations, rehabilitation, and gait analysis.

• Journal of Ocean Engineering and Technology
• /
• v.18 no.4 s.59
• /
• pp.33-39
• /
• 2004

This paper presents a hybrid underwater navigation system for unmanned underwater vehicles, using an additional range sonar, where the navigation system is based on inertial and Doppler velocity sensors. Conventional underwater navigation systems are generally based on an inertial measurement unit (IMU) and a Doppler velocity log (DVL), accompanying a magnetic compass and a depth sensor. Although the conventional navigation systems update the bias errors of inertial sensors and the scale effects of DVL, the estimated position slowly drifts as time passes. This paper proposes a measurement model that uses the range sonar to improve the performance of the IMU-DVL navigation system, for extended operation of underwater vehicles. The proposed navigation model includes the bias errors of IMU, the scale effects of VL, and the bias error of the range sonar. An extended Kalman filter was adopted to propagate the error covariance, to update the measurement errors, and to correct the state equation, when the external measurements are available. To illustrate the effectiveness of the hybrid navigation system, simulations were conducted with the 6-d.o.f. equations of motion of an AUV in lawn-mowing survey mode.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.14 no.5
• /
• pp.423-430
• /
• 2009

Recently as the inactive response characteristics of the existing RCD used on low voltage power distribution system, so control of overload and electric short circuit faults, major causes of electrical fires, are not enough. Therefore, this paper confirms the unreliability of the existing RCD by electrical fault simulator and proposes a EFDPD by using semiconductor switching devices and a high precision current sensor (namely, reed switch) for the prevention of electrical disasters in low voltage power distribution system caused by overload or electric short circuit faults. The sensitive reed switch in the proposed EFDPD exactly detects the increased magnetic flux with the overload or the short current caused by a number of electrical faults, and the following, the self circuit breaker in EFDPD rapidly cuts off the system. The proposed EFDPD confirms the excellent characteristics in response velocity and accuracy in comparison with the conventional circuit breaker through various operation performance analysis. The proposed EFDPD can also prevent electrical disasters, like as electrical fires, which resulted from the malfunction and inactive response characteristics of the existing RCD.

• Journal of Sensor Science and Technology
• /
• v.8 no.5
• /
• pp.407-413
• /
• 1999

The resonant frequency of dielectric rod resonator can be calculated very accurately by means of the concept of electric and magnetic walls from the symmetry. But in a real situation, when the supporter is placed in the cavity, asymmetry appears due to the supporter and its dielectric constant. Then we need research into the resonance characteristic of asymmetric dielectric resonator. In this paper, for the dielectric rod resonator which is placed in the asymmetric position in the conducting cavity, the equation for resonance characteristic was derived and the resonant frequency was calculated with the eigenfunction expansion method. We found that the calculated resonant frequency is very accurate when it was compared with the experimental result and that asymmetry affects the resonant frequency more in TM mode than in TE or in hybrid mode.

• Journal of Sensor Science and Technology
• /
• v.6 no.4
• /
• pp.258-264
• /
• 1997

$YBa_{2}Cu_{3}O_{7}$ single layer dc SQUID magnetometers, prepared on $1\;cm^{2}\;SrTiO_{3}$ substrates, have been fabricated and characterized. Based on the analytical description, a SQUID magnetometer design having a 8.5 mm pickup coil with 2.6 mm linewidth, and a SQUID inductance Ls = 50 pH with $3\;{\mu}m$ Josephson junctions is presented. The devices showed a maximum modulation voltage depth of $65\;{\mu}V$ and a magnetic field noise of 0.6 pT /$\sqrt{Hz}$ at 1 Hz. Clear traces of human magnetocardiogram could be obtained with the SQUID magnetometer operating at 77 K.

• Journal of IKEEE
• /
• v.22 no.4
• /
• pp.1068-1072
• /
• 2018

In underwater communications, there are many challenges due to energy limitations, long propagation delay, low data rate, and high power loss, unlike terrestrial RF communications. Especially, the propagation delay of underwater acoustic channel is five orders of magnitude higher than in electro-magnetic terrestrial channels due to the low speed of sound(1,500m/s). Thus, the MAC protocols for terrestrial communications are not suitable for underwater network. In this paper, we studied the considerations for MAC protocol in underwater acoustic channel. Here, we concentrated on CSMA based MAC protocols. From the results, we confirmed that the number of control packets has an important effect on the performance in underwater environment. These results would be useful in designing MAC protocols for underwater acoustic communications.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.17 no.6
• /
• pp.1275-1282
• /
• 2022

In this study, real-time character navigation using AR lens developed by Nreal is developed. Real-time character navigation is not possible with general marker-based AR because NPC characters must guide while moving in an unspecified space. To replace this, a markerless AR system was developed using Digital Twin technology. Existing markerless AR is operated based on hardware such as GPS, gyroscope, and magnetic sensor, so location accuracy is low and processing time in the system is long, resulting in low reliability in real-time AR environment. In order to solve this problem, using the SLAM technique to construct a space into a 3D object and to construct a markerless AR based on point location, AR can be implemented without any hardware intervention in a real-time AR environment. This real-time AR environment configuration made it possible to implement a navigation system using characters in tourist attractions such as Suncheon Bay Garden and Suncheon Drama Filming Site.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.17 no.2
• /
• pp.325-332
• /
• 2022

In this study, real-time character navigation using AR lens developed by Nreal is developed. Real-time character navigation is not possible with general marker-based AR because NPC characters must guide while moving in an unspecified space. To replace this, a markerless AR system was developed using Digital Twin technology. Existing markerless AR is operated based on hardware such as GPS, gyroscope, and magnetic sensor, so location accuracy is low and processing time in the system is long, which results low reliability in real-time AR environment. In order to solve this problem, using the SLAM technique to construct a space into a 3D object and to construct a markerless AR based on point location, AR can be implemented without any hardware intervention in a real-time AR environment. This real-time AR environment configuration made it possible to implement a navigation system using characters in tourist attractions such as Suncheon Bay Garden and Suncheon Drama Filming Site.

• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.163-177
• /
• 2019

As smartphones are getting widely used, human activity recognition (HAR) tasks for recognizing personal activities of smartphone users with multimodal data have been actively studied recently. The research area is expanding from the recognition of the simple body movement of an individual user to the recognition of low-level behavior and high-level behavior. However, HAR tasks for recognizing interaction behavior with other people, such as whether the user is accompanying or communicating with someone else, have gotten less attention so far. And previous research for recognizing interaction behavior has usually depended on audio, Bluetooth, and Wi-Fi sensors, which are vulnerable to privacy issues and require much time to collect enough data. Whereas physical sensors including accelerometer, magnetic field and gyroscope sensors are less vulnerable to privacy issues and can collect a large amount of data within a short time. In this paper, a method for detecting accompanying status based on deep learning model by only using multimodal physical sensor data, such as an accelerometer, magnetic field and gyroscope, was proposed. The accompanying status was defined as a redefinition of a part of the user interaction behavior, including whether the user is accompanying with an acquaintance at a close distance and the user is actively communicating with the acquaintance. A framework based on convolutional neural networks (CNN) and long short-term memory (LSTM) recurrent networks for classifying accompanying and conversation was proposed. First, a data preprocessing method which consists of time synchronization of multimodal data from different physical sensors, data normalization and sequence data generation was introduced. We applied the nearest interpolation to synchronize the time of collected data from different sensors. Normalization was performed for each x, y, z axis value of the sensor data, and the sequence data was generated according to the sliding window method. Then, the sequence data became the input for CNN, where feature maps representing local dependencies of the original sequence are extracted. The CNN consisted of 3 convolutional layers and did not have a pooling layer to maintain the temporal information of the sequence data. Next, LSTM recurrent networks received the feature maps, learned long-term dependencies from them and extracted features. The LSTM recurrent networks consisted of two layers, each with 128 cells. Finally, the extracted features were used for classification by softmax classifier. The loss function of the model was cross entropy function and the weights of the model were randomly initialized on a normal distribution with an average of 0 and a standard deviation of 0.1. The model was trained using adaptive moment estimation (ADAM) optimization algorithm and the mini batch size was set to 128. We applied dropout to input values of the LSTM recurrent networks to prevent overfitting. The initial learning rate was set to 0.001, and it decreased exponentially by 0.99 at the end of each epoch training. An Android smartphone application was developed and released to collect data. We collected smartphone data for a total of 18 subjects. Using the data, the model classified accompanying and conversation by 98.74% and 98.83% accuracy each. Both the F1 score and accuracy of the model were higher than the F1 score and accuracy of the majority vote classifier, support vector machine, and deep recurrent neural network. In the future research, we will focus on more rigorous multimodal sensor data synchronization methods that minimize the time stamp differences. In addition, we will further study transfer learning method that enables transfer of trained models tailored to the training data to the evaluation data that follows a different distribution. It is expected that a model capable of exhibiting robust recognition performance against changes in data that is not considered in the model learning stage will be obtained.