Title: The Development of an Artificial Neural Networks Aided Image Localization Scheme for Indoor Navigation Applications with Floor Plans Built by Multi-platform Mobile Mapping Systems
Author(s): Jhen-Kai Liao, Guang-Je Tsai
Published in: Proceedings of the 30th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS+ 2017)
September 25 - 29, 2017
Oregon Convention Center
Portland, Oregon
Pages: 3010 - 3027
Cite this article: Liao, Jhen-Kai, Tsai, Guang-Je, "The Development of an Artificial Neural Networks Aided Image Localization Scheme for Indoor Navigation Applications with Floor Plans Built by Multi-platform Mobile Mapping Systems," Proceedings of the 30th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS+ 2017), Portland, Oregon, September 2017, pp. 3010-3027.
Full Paper: ION Members/Non-Members: 1 Download Credit
Sign In
Abstract: *ION GNSS+ 2017 Student Paper Award Winner* Indoor navigation and mapping is popular because of the indispensable smartphone in our daily life. Among all the indoor navigation techniques Pedestrian Dead Reckoning (PDR) has the most potential to confront the challenges of a GNSS-denied environment based on the various embedded sensors. However, PDR has inherent time accumulated errors and aided algorithms such as external infrastructure, frequently stable update and map information, are needed to maintain acceptable result. Another option is to use the image-based localization which detects georeferenced markers in the image to estimate the camera’s position. For this technique, the automation and fast implementation become important. The proposed indoor navigation system is based on the embedded sensors and the integration of two technologies: Artificial Neural Networks (ANN) aided image-based localization and PDR. The distributed georeferenced markers and floor plan are produced by joint Mobile Mapping Systems (MMSs) first. Then, ANN is novel applied to estimate the distance between the marker and camera in real-time. Finally, the camera position is updated through the detected georeferenced marker, estimated distance and orientation from inertial sensor on real time in order to maintain the stable indoor positioning. Compare to the traditional marker-based localization and PDR, the proposed ANN aided image-based localization has better performance, less computational burden and more effective range of marker detection. Meanwhile, the proposed integrated system mitigates the challenges of PDR and image-based localization when they are used independently. The result shows the proposed system is able to initialize in indoor without manually given initial position and provides long-term accurate indoor localization without any infrastructures. In addition, the use of the joint operation of different MMSs for the necessary information collection is flexible, low labor and fast implementation. Especially, the accurate floor plan is the cornerstone of the indoor navigation whether it is for map aided algorithm or presentation.