In the last decade, With the spread of smartphones, indoor positioning technology using sensors has made a lot of progress. Light, Wi-Fi, BLE, methods for estimating landmarks using magnetic fields, and techniques for recognizing and estimating pedestrian activity are various. However, location esti...
In the last decade, With the spread of smartphones, indoor positioning technology using sensors has made a lot of progress. Light, Wi-Fi, BLE, methods for estimating landmarks using magnetic fields, and techniques for recognizing and estimating pedestrian activity are various. However, location estimation techniques using wireless infrastructure or magnetic fields do not take into account the mobility of pedestrians, making it difficult to apply them on the move. In addition, the technology to estimate the location of a smartphone or wearable device by attaching it to a specific part of the body is difficult to apply to ordinary pedestrians because it does not take into account the fact that a pedestrian estimates the location while looking at the smartphone.
Thus, the proposed method is designed to operate while the actual pedestrian is looking at the smartphone, so that the pedestrian's position is measured by the smartphone located at the pedestrian's bust level. The data is collected using smartphone built-in accelerators, gyroscopes, magnetic field sensors and Wi-Fi modules, and is learned through deep learning models such as 1D CNN, LSTM and BLSTM based on the data collected. In this paper, the accuracy of the six activity patterns of pedestrians was 95.2%, the accuracy of
the five walking distance was 96.76%, and the accuracy of the landmark was 97.62%. Therefore, the indoor position is estimated based on six patterns of activity of pedestrians and five walking widths, and cumulative errors due to PDR technique are corrected using the landmarks using Wi-Fi and magnetic field sensors. Thus, by using the proposed method, continuous distance differences in existing PDR systems can be corrected to
reduce indoor positional estimation errors.
In the last decade, With the spread of smartphones, indoor positioning technology using sensors has made a lot of progress. Light, Wi-Fi, BLE, methods for estimating landmarks using magnetic fields, and techniques for recognizing and estimating pedestrian activity are various. However, location estimation techniques using wireless infrastructure or magnetic fields do not take into account the mobility of pedestrians, making it difficult to apply them on the move. In addition, the technology to estimate the location of a smartphone or wearable device by attaching it to a specific part of the body is difficult to apply to ordinary pedestrians because it does not take into account the fact that a pedestrian estimates the location while looking at the smartphone.
Thus, the proposed method is designed to operate while the actual pedestrian is looking at the smartphone, so that the pedestrian's position is measured by the smartphone located at the pedestrian's bust level. The data is collected using smartphone built-in accelerators, gyroscopes, magnetic field sensors and Wi-Fi modules, and is learned through deep learning models such as 1D CNN, LSTM and BLSTM based on the data collected. In this paper, the accuracy of the six activity patterns of pedestrians was 95.2%, the accuracy of
the five walking distance was 96.76%, and the accuracy of the landmark was 97.62%. Therefore, the indoor position is estimated based on six patterns of activity of pedestrians and five walking widths, and cumulative errors due to PDR technique are corrected using the landmarks using Wi-Fi and magnetic field sensors. Thus, by using the proposed method, continuous distance differences in existing PDR systems can be corrected to
reduce indoor positional estimation errors.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.