• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.36 no.5
• /
• pp.395-401
• /
• 2018

Recently, most of mobile devices are equipped with GNSS (Global Navigation Satellite System). When the GNSS signal is available, it is easy to obtain position information. However, GNSS is not suitable solution for indoor localization, since the signals are normally not reachable inside buildings. A wide varieties of technology have been developed as a solution for indoor localization such as Wi-Fi, beacons, and inertial sensor. With the increased sensor combinations in mobile devices, mobile devices also became feasible to provide a solution, which based on PDR (Pedestrian Dead Reckoning) method. In this study, we utilized the combination of three sensors equipped in mobile devices including accelerometer, digital compass, and gyroscope and applied three representative PDR methods. The proposed methods are done in three stages; step detection, step length estimation, and heading determination and the final indoor localization result was evaluated with terrestrial LiDAR (Light Detection And Ranging) data obtained in the same test site. By using terrestrial LiDAR data as reference ground truth for PDR in two differently designed experiments, the inaccuracy of PDR methods that could not be found by existing evaluation method could be revealed. The firstexperiment included extreme direction change and combined with similar pace size. Second experiment included smooth direction change and irregular step length. In using existing evaluation method which only checks traveled distance, The results of two experiments showed the mean percentage error of traveled distance estimation resulted from three different algorithms ranging from 0.028 % to 2.825% in the first experiment and 0.035% to 2.282% in second experiment, which makes it to be seen accurately estimated. However, by using the evaluation method utilizing terrestrial LiDAR data, the performance of PDR methods emerged to be inaccurate. In the firstexperiment, the RMSEs (Root Mean Square Errors) of x direction and y direction were 0.48 m and 0.41 m with combination of the best available algorithm. However, the RMSEs of x direction and y direction were 1.29 m and 3.13 m in the second experiment. The new evaluation result reveals that the PDR methods were not effective enough to find out exact pedestrian position information opposed to the result from existing evaluation method.

• Journal of Internet of Things and Convergence
• /
• v.10 no.4
• /
• pp.101-108
• /
• 2024

Indoor localization is a critical component for numerous applications, ranging from navigation in large buildings to emergency response. This paper presents an enhanced Pedestrian Dead Reckoning (PDR) scheme using smartphone sensors, integrating neural network-aided motion recognition, Kalman filter-based error correction, and multi-sensor data fusion. The proposed system leverages data from the accelerometer, magnetometer, gyroscope, and barometer to accurately estimate a user's position and orientation. A neural network processes sensor data to classify motion modes and provide real-time adjustments to stride length and heading calculations. The Kalman filter further refines these estimates, reducing cumulative errors and drift. Experimental results, collected using a smartphone across various floors of University, demonstrate the scheme's ability to accurately track vertical movements and changes in heading direction. Comparative analyses show that the proposed CNN-LSTM model outperforms conventional CNN and Deep CNN models in angle prediction. Additionally, the integration of barometric pressure data enables precise floor level detection, enhancing the system's robustness in multi-story environments. Proposed comprehensive approach significantly improves the accuracy and reliability of indoor localization, making it viable for real-world applications.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2015.05a
• /
• pp.548-549
• /
• 2015

무인 잠수정은 자율 무인잠수정(이하 'AUV' 또는 '자율무인잠수정'을 혼용)과 원격조정잠수정(이하 'ROV'로 지칭)으로 분류를 할 수 있다. ROV는 테더 게이블로 인한 작업 범위의 한계와 운동성능 효율이 떨어지는 단점을 지니고 있어, 테더 케이블이 필요 없는 AUV에 대한 필요성이 증대되고 있다. 추측 항법 시스템인 관성 항법 시스템(inertial navigation system, 이하 'INS'로 지칭)은 외부 도움없이 관성측정 장치(inertial measurement unit, 이하 'IMU'로 지칭)를 활용하여 구성된 시스템을 말한다. IMU는 자이로 스코프(gyroscope), 가속도계(accelerometer), 지자기(magnetic)센서로 구성된 측정 장치로 3개의 센서를 사용하여 상호 보정을 통한 기동 체의 위치, 속도 및 자세 정보를 제공한다. 복합항법시스템은 추측항법시스템이 가지는 누적오차와 측위 항법시스템이 가지는 외부환경에 대한 단점을 상호 보완하는 방법으로 연구가 진행 중이다. 하지만 심해서 또는 해양의 특성에 따라 측위 시스템이 사용되지 못하기 때문에 추측 항법시스템의 다양한 관성 센서를 활용한 상로 보완과 신호처리 방법을 통한 연구 개발이 진행 중이다. 다양한 센서 정보를 통합하는 목적으로 칼만 필터와 같은 최적 필터기법이 보편적으로 사용되고 있다. 칼만 필터는 확률 선형 시스템에 대하여 공정잡음 및 측정 잡음이 가우시안 확률 분포를 따를 때 최적의 추정자가 된다. 또한 가우시안 조건을 만족하지 않는 경우에도 선형 추정자 중에 추정 오차의 분산이 가장 작은 추정자이다. 칼만 필터가 최상의 성능을 발휘 하려면 공정잡음과 측정 잡음의 실제 값을 정확히 알아내는 것이 중요하다. 잡음 수준에 대한 정보가 부정확 할 경우 칼만 필터는 발산 할 수 있기 때문에 시스템에서 잡음 수준의 공산은 칼만 필터의 최적 이득을 결정하는 중요한 요소로 추정치에 큰 영향을 준다. 따라서 칼만 필터를 추측항법시스템에 적용 시킬 경우 실제 모텔의 잡음 공분산을 정확히 추정할 수 있는 기법이 요구된다. 추측항법시스템은 다양한 센서를 활용하기 때문에 움직이는 기동 표적에 적용시 잡음공분상이 변하기 때문에 항법시스템이 저하 될 수 있다. 본 연구에서는 다양한 센서를 융합하여 해양 생체 로봇의 정밀 자세 제어가 가능한 시스템을 제안하고자 한다.

• 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.

• KIPS Transactions on Computer and Communication Systems
• /
• v.12 no.5
• /
• pp.171-180
• /
• 2023

The three major sleep disorders in Korea are snoring, sleep apnea, and insomnia. Lack of sleep is the root of all diseases. Some of the most serious potential problems associated with sleep deprivation are cardiovascular problems, cognitive impairment, obesity, diabetes, colitis, prostate cancer, etc. To solve these problems, the Korean government provided low-cost national health insurance benefits for polysomnography tests in July 2018. However, insomnia patients still have problems getting treated in terms of time, space, and economic perspectives. Therefore, it would be better for insomnia patients to be allowed to test at home. The measuring device can measure six biosignals (eye movement, tossing and turning, body temperature, oxygen saturation, heart rate, and audio). A gyroscope sensor (MPU9250, InvenSense, USA) was used for eye movement, tossing, and turning. The input range of the sensor was in 258°/sec to 460°/sec, and the data range was in the input range. Body temperature, oxygen saturation range, and heart rate were measured by a sensor (MAX30102, Analog Devices, USA). The body temperature was measured in 30 ℃ to 45 ℃, and the oxygen saturation range was 0% for the unused state and 20 % to 90 % for the used state. The heart rate measurement range was in 40 bpm to 180 bpm. The measurement of audio signal was performed by an audio sensor (AMM2742-T-R, PUIaudio, USA). The was -42 dB ±1 dB frequency range was 20 Hz to 20 kHz. The measured data was successfully received in wireless network conditions. The system configuration was consisted of a PC and a mobile app for bio-signal measurement and data collection. The measured data was collected by mobile phones and desktops. The data collected can be used as preliminary data to determine the stage of sleep and perform the screening function for sleep induction and sleep disturbances. In the future, this convenient sleep measurement device could be beneficial for treating insomnia.