Enhanced PDR Technology Based on Motion Grouping using Smartphone IMU

Other Titles
스마트폰 내 IMU 센서를 활용한 Motion Grouping 기반 enhanced PDR 기술 개발
Issue Date
2020년 항법시스템학회 정기학술대회
In this paper, we propose a technology to enhance the performance of Pedestrian Dead-Reckoning (PDR) based on motion grouping of a user who own smart phone. Previously, there were mainly studies for PDR using the Inertial Measurement Unit (IMU) in a smart phone. However, there is a problem in that the accuracy of the PDR is deteriorated by identifying movements other than walking when a user makes unspecified movements, such as when a user looks around or walks with a different location of a smartphone. Thus, the proposed technology identifies defined motion groups using pattern of data about acceleration and attitude (roll or pitch) changing by position of smartphone for reducing error due to unspecific movements. The motion group is determined by applying peak detection algorithm to each of the acceleration and attitude information. Step and heading information are determined according to the identified motion group. In addition, the consistency analysis of the detected peaks improves the accuracy of the step detection and step length estimation by removing steps in unspecified motions. This eventually leads to improved performance of the PDR. To verify the performance of the proposed technology, we performed a test at the Lotte department store in Myeong-dong. Like real visitor to a department store, we conducted the test on various motions such as looking around a store or putting a smartphone in a pocket. The test results show 90% accuracy in step detection using only accelerometer. Otherwise, the proposed technology shows 97% accuracy in step detection. It was confirmed the PDR was improved due to the removal of the step detection error caused by unspecific movements. Through the results, we demonstrate the proposed technology
smartphone; motion grouping; step detection; PDR
Appears in Collections:
KIST Conference Paper > 2020
Files in This Item:
There are no files associated with this item.
RIS (EndNote)
XLS (Excel)


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.