Search

There are some constraints such as external electrodes, a failure to capture most paroxysmal atrial fibrillation (AFib), low power transfer efficiency (PTE) for 24/7 charging technology, a short period of monitoring, and automatic detection of AFib in conventional electrocardiogram (ECG) sensors. To overcome these constraints, an implantable ECG sensor with a 2-coil inductive link with maximum power transfer efficiency (PTE) is designed to continuously monitor patients and efficiently detect AFib using global covering rule discovery and the minimum description length (MDL) algorithm. Among different combinations of ECG coils, the square spiral-square spiral coil demonstrates the maximum PTE, 56.23%, at the resonant frequency of 13.56 MHz and it is used in the implantable ECG sensor. The QRS complex from ECG signals of twenty-nine AFib patients is detected using different operation methods (DOM). The MDL algorithm is used to group 12 features of heart rate variability (HRV) parameters. The global covering rule discovery is proposed as a novel classification technique of AFib in ECG data. The average classification accuracy was 96.67 ± 7.03, and then the average recall, precision, F1-measures, and an average number of generated rules were 97.08 ± 6.23, 97.08 ± 6.23, 96.57 ± 7.23, and 7.9 ± 0.32, respectively. We found that the NN50, pNN50, and LF parameters can distinguish the AFib patient better than a healthy one. Among these parameters, pNN50 showed that it is greater than 34.75 in 41.38% of patients. The optimized implantable ECG sensor with a maximum PTE of 56.23% along with novel AFib detection and classification methods is suitable for its implementation in future implantable ECG sensors.

Deep neural networks (DNN) have been applied to numerous artificial-intelligence applications because of their remarkable accuracy. However, computational requirements for deep neural networks are recently skyrocketing far beyond the Moore’s Law. In addition to the importance of accuracy, the industry’s demand for efficiency in model learning process is increasing. This has led to various attempts to make DNNs more lightweight. Hence, we propose a modeling technique that applies lightweight convolutional neural networks (CNN) to handle the model-learning processes for DNNs. The proposed spatial-shift pointwise quantization (SSPQ) model elegantly combines compact network-design techniques to revitalize DNN quantization efficiency with little accuracy loss. We set the depths of our SSPQ model to 20, 34, and 50 to test against CIFAR10, CIFAR100, and ImageNet datasets, respectively. By applying SSPQ20 to the CIFAR10 dataset, we reduced accuracy degradation by 2.95%, while reducing the number of parameters $8.7\times $ . For the same dataset, our “wide” SSPQ20 variant reduced training parameters $1.96\times $ , compared with the ResNet20 model architecture, which provided a 0.2% improvement. By applying SSPQ34 to the CIFAR100 dataset, we successfully reduced the number of learning parameters $8\times $ , compared with the ResNet34 model with an accuracy degradation of 4.57%. By applying SSPQ50 to the ImageNet dataset, we successfully reduced the number of parameters $10.2\times $ over ResNet50 with an accuracy degradation of only 2.68%. Therefore, using only 9.78MB of learning parameter, the SSPQ50 model guarantees 73.6% accuracy. This is an improved performance by 1.46% in terms of accuracy and 3.62MB in terms of model size compared to the MobileNetV2 lightweight model. The source code is available at https://github.com/Eunhui-Kim/SSPQ .

The Visual Object Tracking challenge VOT2021 is the ninth annual tracker benchmarking activity organized by the VOT initiative. Results of 71 trackers are presented; many are state-of-the-art trackers published at major computer vision conferences or in journals in recent years. The VOT2021 challenge was composed of four sub-challenges focusing on different tracking domains: (i) VOT-ST2021 challenge focused on short-term tracking in RGB, (ii) VOT-RT2021 challenge focused on "real-time" short-term tracking in RGB, (iii) VOT-LT2021 focused on long-term tracking, namely coping with target disappearance and reappearance and (iv) VOT-RGBD2021 challenge focused on long-term tracking in RGB and depth imagery. The VOT-ST2021 dataset was refreshed, while VOT-RGBD2021 introduces a training dataset and sequestered dataset for winner identification. The source code for most of the trackers, the datasets, the evaluation kit and the results along with the source code for most trackers are publicly available at the challenge website 1 .

3D Multi-stacking technology using Cu-Cu hybrid wafer bonding has been developed to achieve superior power, speed performances and higher density with minimized form factor. To realize multi-stacked device by using Wafer on wafer (WoW) bonding, both Face-to-Face (F2F) and Back-to-Face (B2F) interconnection can be required for more efficient and simple I/O pad design as well as lower resistance thanks to shorter interconnections. In this paper, we successfully demonstrated wafer-level multi-stacking structure by applying robust Cu-to-Cu bonding combined with Through-Si-Via (TSV) process. Three key subjects of comprehensive understanding about bonding interface, TSV and edge engineering to avoid defects and yield drop were essential to fulfill this research. Chemical-Mechanical-Polishing (CMP) and surface treatment had a huge impact on the Cu/dielectric hybrid bonding surface. In particular, planarization at the backside of the wafer is the key technology for robust follow up wafer bonding. For TSV formation, etch advancement performed to control the enhanced target compared to existing scheme. One more key subject is the edge engineering including integration considered edge treatment and effect by the process has been conducted to improve wafer bonding quality and reproducibility. Through the optimization of multi-stack oriented processes and integration, we confirmed extremely high yield of electrical connection including wafer edge for multiple layers. Many valuable applications can be introduced for sensors, memories, logic devices and even combinations of them by utilizing this novel processes.

We propose a low-temperature co-fired ceramic (LTCC)-based leadless surface mount package to improve switching and heat-dissipation properties of a power semiconductor. A silicon carbide (SiC) Schottky barrier diode (SBD) bare die is embedded to a cavity in the LTCC-based multi-layer substrate. Instead of conventional aluminum wires, a flat copper clip is used to reduce parasitic inductance and electrical resistance caused by bond wires. Multiple vias filled with silver are embedded in the multi-layer substrate to decrease the electrical resistance and to improve heat-dissipation of the package. The measured reverse recovery charge (Q rr ) was 18.02 nC at a reverse voltage of 300 V and $di/dt$ of $300\ \mathrm{A}/\mu\mathrm{s}$ . It exhibits 18.7% improvement in the Q rr as compared to the conventional TO-220 packaged product using the same bare die. The power loss and heat-dissipation performances of the proposed package was evaluated through a power factor correction (PFC) converter. In comparison with commercial products, we can see that the proposed package has low-loss and high heat-dissipation properties.

A 2.8μm 4-tap global shutter pixel has been realized for a compact and high-resolution time of flight (ToF) CMOS image sensor. 20,000 e- of high full-well capacity (FWC) per a tap is obtained by employing a supplementary MOS capacitor. 36% of high quantum efficiency (QE) has been achieved by backside scattering technology (BST) and thick silicon process. In addition, demodulation contrast (DC) is improved to 86 % by additional deep photodiode doping process for static potential gradient in a photodiode.

We propose a multi-layer ceramic based surface mount device (SMD) packaging for 1200 V and 1700 V silicon carbide (SiC) Schottky barrier diode (SBD) power semiconductors. The distance between anode and cathode on a ceramic substrate is 1.0 mm to achieve an isolation voltage performance over 2000 V. We implemented the proposed multi-layer ceramic based substrates with 8 mm x 8 mm size. Using the substrates, we fabricated three SMD typed power semiconductor packaging such as 1200 V/2 A, 1200 V/5 A, and 1700 V/5 A-class SiC SBD and evaluated forward/reverse static and dynamic performances. It is the world’s first result of SMD packaging using a multi-layer ceramic substrate for 1200 and 1700 V SiC SBD.

In this paper, the multilayer power inductor with the air-clip is designed and fabricated to improve the saturation current of power inductor for the power conversion system. It is proved that the air-clip is able to influence the saturation current of power inductor and the efficiency of the power conversion system. Also, the possibility of the power inductor with the air-clip in the power conversion system for improving rated current and efficiency is demonstrated.

Intersection classification is one of the key components of autonomous navigation. Several related research works have been conducted as a prior task to solve problems such as autonomous driving, aircraft hovering, and navigating in mines. However, to the best of our knowledge, none of these studies support pedestrian-view navigation to guide the small and slow robots for which it is too dangerous to be operated on normal roads along with normal vehicles. To address this problem, we propose: 1) a pedestrian-view-level intersection classification image dataset, 2) ResNet-based architecture fine-tuned on the proposed dataset, and 3) thorough experimentation to explore the capabilities of our proposed architecture. The detailed analysis reported in this paper enabled us to find the network configuration that is neither underfit nor overfit to our data and achieves 80% test accuracy.

Extensive research has been carried out on intersection classification to assist the navigation in autonomous maneuvering of aerial, road, and cave mining vehicles. In contrast, our work tackles intersection classification at pedestrian-view level to support navigation of the slower and smaller robots for which it is too dangerous to steer on a normal road along with the usual vehicles. Particularly, we focus on investigating the kind of features a network may exploit in order to classify intersection at pedestrian-view. To this end, two sets of experiments have been conducted using an ImageNet-pretrained ResNet-18 architecture fine-tuned on our image-level pedestrian-view intersection classification dataset. First, ablation study is performed on layer depth to evaluate the importance of high-level feature, which demonstrated superiority in using all of the layers by yielding 77.56% accuracy. Second, to further clarify the need of such high level features, Class Activation Map (CAM) is applied to visualize the parts of an image that affect the most on a given prediction. The visualization justifies the high accuracy of an all-layers network.

In this paper, a power inductor embedded pattern with multi-via is proposed to achieve the lower DCR. Also, the low temperature co-fired ceramic (LTCC) substrate for the power conversion module was fabricated to be interconnected with a proposed power inductor. A proposed power inductor can be interconnected to the bottom side of the LTCC substrate. Two types of power inductors were fabricated to compare electrical properties. One is with the single-stack multi-via pattern (SMP) and the other is with the dual-stack multi-via pattern (DMP). The DCR, inductance at the frequency of 1 MHz, and Q-factor at the frequency of 1 MHz for the fabricated inductor with SMP were 17.82 mΩ, 3.21 μH and 42.19, respectively. On the other hand, that of the fabricated inductor with DMP were 12.49 mΩ, 3.3 μH, and 56.77, respectively. Compared to the fabricated inductor with SMP, the inductor with DMP shows lower DCR and higher Q-factor while having a similar inductance. A LTCC based DC-DC converter as the power conversion module was designed to be interconnected with proposed inductors. A DC-DC converter with switching frequency of 1 MHz has input voltage of 12 V, output voltage of 5 V, and rated power of 35 W. The efficiency of the converter with SMP inductor at 35 W was 92.41 % and that of the converter with DMP inductor was 93.79 %. A proposed inductor was designed to minimize DCR and to improve the stability of fabrication. Based on the measured results, it is demonstrated that the DCR is decreased by a proposed structure. This is due to the multi-via pattern in the proposed inductor. It is also proved that a proposed inductor is very suitable for improving the efficiency of the power conversion module.

Data augmentation has been proven effective which, by preventing overfitting, not only enhances the performance of a deep neural network but also leads to a better generalization even with limited dataset. Recently introduced regional dropout based data augmentation strategies remove (or replace) some parts of an input image with a desideratum to make the network focus on less discriminative portions of an image, which results in an improved performance. However, such approaches usually possess' strong-edge' problem caused by an obvious change in the pixels at the positions where the image is manipulated. It may not only impact on the local convolution operation but can also provide clues for the network to latch on to, which do not align well with the fundamental philosophy of augmentation. In order to minimize such peculiarities, we introduce Smoothmix in which blending of images is done based on soft edges and the training labels are computed accordingly. Extensive analysis performed on CIFAR-10, CIFAR- 100 and ImageNet for image classification demonstrates state-of-the-art results. Furthermore, Smoothmix significantly increases the robustness of a network against image corruption which is validated by the experiments carried out on CIFAR-100-C & ImageNet-C corruption datasets.

In an ambient intelligence (AmI) environment, electronic devices that comprise the Internet of things (IoT) network work together seamlessly to provide a wide variety of applications and intelligent services to users. Computer-assisted pronunciation training (CAPT), a widely used application in the traditional Internet environment that corrects user’s pronunciation, is a promising service for transition to the AmI environment. However, the migration of the CAPT to the AmI environment is challenging due to its high computational requirements that is at odds with the low computational capacity of IoT members. In this paper, we propose a smartphone-assisted pronunciation learning technique based on a lightweight word recommendation method that exploits built-in functions supported by IoT members and a computationally moderate word selection method. The experimental evaluation of the proposed method demonstrates that the user pronunciation is significantly improved without incurring unacceptable computational costs for a smartphone platform.

we will introduce a new program so-called TVA(Tech-Venture Academy) for the setup of a role model, and cultivation of outstanding enterprise innovation experts and investigate the performance of its program to overcome the crisis faced by the Korean manufacturing industry due to the global economic recession and the dumping of companies in developing countries, and to lead the role of regional industry promotion and also, to introduce innovation management experts and industry re-creation because DGIST should play as a science and technology specialization and leading university located in Daegu of South Korea.

The new generation of consumer wristbands are now able to classify sleep stages based on multiple sensing data. Several studies have validated the accuracy of one of the latest models, that is, Fitbit Charge 2, in measuring polysomnographic(PSG) parameters. Nevertheless, its feasibility in measuring sleep disorder remains unknown. This study aimed to evaluate the potential of Fitbit Charge 2 as a means of monitoring sleep disorders for workers. We recruited participants without mental and medical diseases from 3 sites consisting of department, hospital and research institute. Sleep data were obtained from 3-month night recording at each worker's home in 95 adults (22-64 years; 68 women; over 20 participants from each site). Our results indicate that there are some differences in sleep patterns in different occupations.

The antenna system of the active phased array radar is composed of parallel transmission and reception paths. The transmission and reception paths of the parallel structure include nonlinear errors. And a long time test measurement is performed using a large measuring facility to align them electrically. If components in parallel path are replaced, a radar system has to calibrate a difference of characteristic to maintain the alignment as confirmed by the test before. In the case of an array radar system with many transmitted receiver module(TRM)s, an efficient calibration algorithm is required, which takes into account operating time and initial alignment error. Especially in the case of TRM, which is designed as an integrated antenna considering the maintenance property, The transmission characteristics of the TRM can be measured, and the measured data can be utilized for the calibration of the antenna system. In this antenna system calibration method, the measuring equipment should be managed, because the influence of the characteristics of the measuring equipment on the compensation state should be minimized. In this paper, we explain a measurement method for the characteristics of a TRM designed as an integrated antenna, and propose a method to compensate the characteristics of measuring equipment. And the effectiveness of the proposed calibration method is verified through simulation.

User specific preamble transmission and timing advance command mechanisms are widely adopted for cellular uplink synchronization. However, the signaling overheads are scaled with the number of users and become prohibitive in massive connectivity usage scenario. In this paper, a kind of cooperative synchronization method is discussed resorting to cell specific signals. More specifically, the cluster of synchronous cells transmit time advanced signals cooperatively, and users estimate individual timing advances based on the reception time.

With the recent advancements in the field of explainable artificial intelligence (XAI), which covers the domain of turning deep network architectures from black boxes to comprehensible structures, it became easy to understand what goes on inside a network when it predicts an output. Many researchers have successfully shown the ‘thought process’ behind a network’s decision making. This rich and interesting information has not been utilized beyond the scope of visualizations once the training finish. In this work, a novel idea to utilize this insight to the network as a training parameter is proposed. Layer-wise Relevance Propagation (LRP), which obtains the effect of each neuron towards the output of the whole network, is used as a parameter, along with learning rate and network weights, to optimize the training. Various intuitive formulations have been proposed, and the results of the experiments on MNIST and CIFAR-10 datasets have been reported in this paper. Our proposed methodologies show better or comparable performances against conventional optimization algorithms. This would open a new dimension of research to explore the possibility of using XAI in optimizing the training of neural networks.

Anomaly detection has stayed among the core topics of research in machine learning and computer vision for decades. Due to its complex nature, limitless possibilities of abnormal cases and the bounded amount of normal examples available, it is still among the popular topics of research. With the increase in surveillance cameras, it is becoming important to have systems which can learn from the vastly available normal data to identify any unusual happenings in the videos. In this paper, an attempt has been made to evaluate and compare the recently introduced such systems. In order to evaluate the methods for implementation as well as their possible applications, we have classified the major papers introduced in last 5 years by the network architectures and the datasets that are used to evaluate these methodologies.

We propose an appropriate decoding algorithm based on ordered statistics for ultra-reliable low latency communication (URLCC) use cases, which typically require low code rates and short block length. This paper investigates tradeoff between performance and complexity of ordered statistics decoding (OSD) algorithms for BCH codes with low code rates over AWGN channel. The block error rate (BLER) performance of the proposed algorithm approaches that of the original OSD resulting in lower complexity. In addition, it outperforms the segmentation-based OSD at a cost of slightly higher complexity.

Large scale connection is one of the main goals of 5G mobile communication. The NB-IoT system achieves this goal using only very narrow bandwidth. NB-IoT systems can also work together without affecting existing LTE systems. In this paper, we compared the inband-mode and the guardband-mode of the NB-IoT system using the BER performance curve.

In wireless communications, orthogonal waveforms having short duration are desirable to achieve low system latency. In addition, immunity to interferences from other user signals is one of desirable features for the waveforms especially in massive uplink access. In this paper, we derive sufficient constraints for orthogonal and short duration waveforms. Under the constraints, we propose signal waveforms suitable for quasi-synchronized communication systems by using gradient-decent algorithm in minimizing average power of interference from other users. Numerical results show that signal-to-interference power ratio can be enhanced by using the proposed waveforms.

We implemented NB- IoT base station and terminal based on the 3GPP Release 13. Our NB- IoT system has been applied to the Smart Factory and the railroad monitoring system. In this paper, we introduce the AtoN NB-IoT system which applied the NB-IoT technology to the AtoN facility.

This paper presents considerations for designing physical random access channel in 5G new radio standard defined in 3GPP. We carried out fixed point link level simulation to see the practical performance in designing modem chip for UE and gNB. In the fixed point simulation, we assumed the sampling rate of 61.44 Msps, 30 kHz subcarrier spacing, 3.7 GHz center frequency, and 40 MHz system bandwidth. To reduce the complexity in the receiver, we employed the alpha max plus beta min algorithm, which is a highspeed approximation of the square root of the sum of two squares. We refers to the performance requirements described in 3GPP TS38.104, which describes the official performance requirements for physical random access channel for new radio. Simulation results indicates that the performance of the designed PRACH module suffices the requirements defined in release 15 at least in additive white Gaussian noise and fading channel with 400 Hz frequency offset.

We propose a 12-to-5 V DC/DC converter using ceramic based multi-layer circuit technologies. By using eight metal layers and embedded vias in seven ceramic layers, the proposed converter can minimize the power losses caused by the power and ground patterns and the noise caused by control lines. The measured efficiency of the proposed converter is 95.97% with a supply voltage of 12 V and output power of 35 W. The line regulation is 0.016% with ±10% change of input voltage of 12 V. The load regulation is 0.54% with 10–100% load change. The dynamic load regulation is 4.78% with 50-100% transition. The maximum ripple and noise is 48.0 and 352.0 mVpp, respectively. We can confirm that the proposed converter operates very stably with outstanding performances.

The scaling of semiconductor device below 10nm has faced the higher process difficulty and longer development periods. Three-dimensional integrated circuits (3D IC) using chip partitioning and wafer-to-wafer bonding have been acknowledged as the next generation semiconductor stacking technology because of smaller form factor, higher density integration and higher performance compared to same-node devices. Wafer-to-wafer bonding is widely used in stacked CMOS image sensor, that is, the bonding between pixel and logic wafer, and this technology has the potential to apply other semiconductor devices. Cu-Cu hybrid bonding has achieved by simultaneous wafer bonding of metal (Cu-Cu) and dielectric materials. In this study, it is investigated on the microstructure of Cu pad for Cu-Cu bonding after post-electroplating and post-bonding annealing process. The Cu grain size distribution and orientation are analyzed with different anneal temperature, which is applied on electroplated Cu, and with additional heat treatment as post-bonding process. The effect of pad size as well as the position within pattern array on Cu microstructure is also studied as the bonding pad is required smaller and smaller size for high density bonding. After Cu-Cu bonding, the cross-section analysis of bonding interface is carried out to see inter-diffusion of Cu atoms across the opposite Cu pad. Cu-Cu hybrid bonding is applied to test vehicle having the daisy chain of 2.4 million. The electrical resistance is measured before and after thermal stress and the Cu-Cu bonding interface is confirmed as robust structure.

This paper presents performance evaluation results of physical random access channel in 5G new radio standard defined in 3GPP, which was derived from simulation assumptions such as preamble formats, wireless channel, signal-to-noise ratio. We assumed sampling rate of 61.44 Msps, 30 kHz subcarrier spacing, 3.7 GHz center frequency, and 40 MHz system bandwidth in the simulation. As the official performance requirements for new radio physical random access channel is not available yet at the moment, which is being discussed in the 3GPP meeting and by email, this paper refers to the requirements described in 3GPP release-13. Simulation results indicates that the performance of NR PRACH defined in 3GPP release 15 meets the requirements in release 13 at least in additive white Gaussian noise and extended typical urban channel.

In this paper, we introduce the AtoN NB-IoT system which applied the 3GPP NB-IoT technology to the AtoN facility. Using this AtoN NB-IoT service, we can remotely manage marine transport facilities and provide various information for safe navigation to ships and related organizations.

Massive connectivity is one of the major goals of 5G mobile communications. The NB-IoT system achieves this goal using only a very narrow bandwidth. The NB-IoT system can also work together without affecting existing LTE systems. In this paper, we analyze the interferences of guardband NB-IoT system from the LTE system when they coexist.

We experimentally demonstrate application of an etched fiber immersed in liquid mxene as a saturable absorber in a passively mode-locked erbium doped fiber laser. In the experiment, we obtained soliton pulses with pulse duration of ~800 fs at a repetition rate of 6.22 MHz in the anomalous regime (-0.65 ps2) at the wavelength of 1556 nm.

The typical method of monitoring arrhythmia is to use a body patch type sensor with a wet electrode. It has several problems caused by wet electrodes for long-term monitoring. Thus, we develop a Smart Outhdoor Shirts equipped with a dry electrode electrocardiogram (ECG) sensor for a cardiac arrhythmia computer aided diagnosis system in this study. The sensor can be inserted in a console close to the chest, charged, used to communicate wirelessly, and connected with a smartphone application. According to experiments, the ECG signals measured by the smart shirt indicated that 97 $5\pm 1\%$ of the signals could be measured in an immobile state and at least 85. $2\pm 2\%$ of the signals could be measured during movement. In addition, we propose a computer aided diagnosis system for detecting cardiac arrhythmia. It was determined through experiments that the system can detect arrhythmia with an accuracy of 98. $2\pm 2 \%$ .

In the physical human-robot interaction system, improving transparency performance which allows humans to move as if there is no robot is a challenging topic. In general physical human-robot interaction, the multiaxial force sensor is used to control the robot. However the signal measured from the force sensor involves not only the force applied to the robot by the human motion intention but also the influence of the natural force feedback between the robot and the human hand due to the motion of the robot. Therefore, in order to improve the transparency performance, it is necessary to consider the dynamics of the human hand as well as the dynamics of the robot. In this paper, an novel algorithm to improve transparency performance is proposed. The proposed algorithm uses only the multiaxial force sensor and compensates the natural force feedback by using the human hand impedance and further improves the transparency performance which has a limitation adjusting admittance parameters. The validity of the proposed algorithm is described through the system model analysis and the experiment using the hydraulic upper limb exoskeleton.

The present study deals with optimization of machining characteristics using Multi characteristic Optimization in collaboration with utility concept. Surface roughness & Tool wear rate are the selected response characteristics for investigation. The study makes use of Taguchi's robust design experiments with Utility concept. Each experiment was conducted using different combinations of input parameters. The investigations suggest that current plays a significant role in optimizing both the characteristics followed by duty cycle, voltage & Pulse on time. The study also reveals significant interactions between current and pulse on time, current and spark gap & pulse on time and spark gap. The obtained results and the settings of the machine are further validated by the confirmation experiments.

Many researchers have been studying a walking robot that can walk in an Unknown environment. But there is a problem. When the leg is landing, the motor or torque sensor is damaged by the amount of impact generated from the uncleaned ground. In order to solve this problem, many researchers have tried to reduce the amount of impulse by placing a passive compliance element in a motor-pack. The efforts have been made to reduce the impact by creating virtual m, c, and k elements through active compliance, impedance control or admittance control. In this paper, we designed a new concept motor-pack that enters the joint of a walking robot. The main function is to make the motor-pack into a package type to protect important components from external shocks and speed up the communication speed due to the circuit intensiveness. The motor-pack consists of motor, gear, bearing, encoder and torque sensor. In addition, this paper validates the SEA module as a torque sensor.

In this paper, A simple method for estimating the impedance of a human hand is proposed. In a physical human-robot interaction situation, safety is a very important issue because humans are in contact with robots. Since robots become unstable when they contact a stiff external environment or when there is a difference between human motion intention and robot motion, knowing the dynamics of a human hand plays an important role in improving stability and control performance in physical human-robot interaction. The proposed simple method is to drop an object on the human hand and measuring external force and the movement distance of the hand to estimate the impedance parameter. The proposed method can estimate the impedance of the human hand easily and simply than the conventional methods using the robot arm or the dedicated equipment. The validity of the proposed method is verified by comparing simulation results with estimated impedance and actual experimental results.