| Title |
A Dead Reckoning Calibration Approach using Activity Recognition for Indoor Pedestrian Localization |
| Authors |
이혁(Hyuck Yi) ; 김준성(JunSeong Kim) |
| DOI |
https://doi.org/10.5573/ieie.2020.57.1.70 |
| Keywords |
Pedestrian dead reckoning ; Activity recognition ; Feature-point floor plan ; Indoor localization ; |
| Abstract |
Dead reckoning is widely used for localization services with the advantage of costs and toughness to environmental changes. However, it has a critical drawback on errors, which accumulate over time. Inside buildings, there are space that constrain user's movement. Depending on the structure of a building, users perform typical actions, such as turning at intersections and walking through corridors. In this paper, we present a dead reckoning calibration system for indoor pedestrian localization. Define feature-points, where users’ specific behaviors are expected and construct a simple feature-point floor plan of a building. Using inertial sensors we can identify user's activity and can identify the user's position by searching its corresponding feature-point on the simple floor plan. We develop a data logger using ATmega328 processor to validate the calibration system. Participants carrying the device move along a predefined experimental route. Since there are, in general, many feature-points within a building, the error by dead reckoning can be limited to a relatively short interval between those feature-points. The experimental results show that the errors of 11.82m by dead reckoning through the 157.7m of experimental route is spread by 13 feature points and is limitied to 2.02m between them. |