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

• DOI: 10.33012/2017.15212
• Publication Date: 2018-12-11T21:23:22Z
• Authors: Jhen-Kai Liao; Guang-Je Tsai
• Keywords: Dead reckoning; Mobile mapping; Georeference

*ION GNSS+ 2017 Student Paper Award Winner* Indoor navigation and mapping is popular because of the indispensable smartphone in our daily life. Among all indoor techniques Pedestrian Dead Reckoning (PDR) has most potential to confront challenges a GNSS-denied environment based on various embedded sensors. However, PDR inherent time accumulated errors aided algorithms such as external infrastructure, frequently stable update map information, are needed maintain acceptable result. Another option use image-based localization which detects georeferenced markers image estimate camera's position. For this technique, automation fast implementation become important. The proposed system sensors integration two technologies: Artificial Neural Networks (ANN) PDR. distributed floor plan produced by joint Mobile Mapping Systems (MMSs) first. Then, ANN novel applied distance between marker camera real-time. Finally, position updated through detected marker, estimated orientation from inertial sensor real order positioning. Compare traditional marker-based PDR, better performance, less computational burden more effective range detection. Meanwhile, integrated mitigates when they used independently. result shows able initialize without manually given initial provides long-term accurate any infrastructures. In addition, operation different MMSs for necessary information collection flexible, low labor implementation. Especially, cornerstone whether it algorithm or presentation.