다양한 모션에서 AI를 활용한 스마트폰 기반 PDR 기술 향상 연구

With the widespread use of smartphones, there has been a significant advancement in Pedestrian Dead Reckoning (PDR)technology using smartphones equipped with various sensors. This study proposes a method to improve the accuracy of stepdetection for smartphone users by applying Artificial Intelligence (AI). To achieve this, we plan to utilize the accelerometerand gyroscope sensors of smartphones to analyze various user motions and train a deep learning-based model to learn steppatterns. During the research process, diverse motion data from smartphone users will be collected and used to train themodel, which will then be compared and analyzed against various existing step detection algorithms. It is expected that theAI-based step detection model will demonstrate higher accuracy compared to traditional methods. High-accuracy stepdetection can help reduce the accumulation of step errors, which often lead to significant discrepancies in PDR. The resultsof this study are anticipated to enhance the accuracy of the Surface Correlation (SC) technique, which combines PDR withFingerprinting. Additionally, this approach is expected to be beneficial in various fields, including location-based servicesand user behavior analysis.