The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLVI-3/W1-2022
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLVI-3/W1-2022, 67–73, 2022
https://doi.org/10.5194/isprs-archives-XLVI-3-W1-2022-67-2022
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLVI-3/W1-2022, 67–73, 2022
https://doi.org/10.5194/isprs-archives-XLVI-3-W1-2022-67-2022
 
22 Apr 2022
22 Apr 2022

PEDESTRIAN INERTIAL NAVIGATION ALGORITHM BASED ON SCENE RECOGNITION

T. Lei1 and Y. Li2,3 T. Lei and Y. Li
  • 1School of Geodesy and Geomatics, Wuhan University, Luoyu Road, China
  • 2Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Luoyu Road, China
  • 3Hubei Luojia Laboratory, Wuhan, China

Keywords: Pedestrian Navigation, Inertial Navigation, Scene Recognition, EKF

Abstract. As indoor location-based services become increasingly essential to people's daily life, it is necessary to build a stable and accurate indoor pedestrian positioning system. The foot-mounted inertial navigation system can provide a short-term robust solution but suffer from error accumulation over time. To alleviate this issue, this paper proposes a pedestrian inertial navigation algorithm based on scene recognition to reduce the heading drift. Based on the hypothesis that corridors in buildings are generally narrow and straight and pedestrians have a high probability to walk in straight lines in corridors, we use a scene recognition model to assist foot-mounted INS. When the scene recognition model determines that the pedestrian is walking in a corridor, a straight-line constraint will be implemented to reduce the heading drift and improve the observability of the vertical gyroscope bias. Experiments show that the algorithm can effectively improve the navigation performance and the observability of vertical gyroscope bias when the sensor biases are significant.