Title: Inertial Sensors Strapdown Approach for Hybrid Cameras and MEMS Positioning
Author(s): Paolo Dabove, Irene Aicardi, Nives Grasso, Andrea Lingua, Giorgio Ghinamo, Cecilia Corbi
Published in: Proceedings of IEEE/ION PLANS 2016
April 11 - 14, 2016
Hyatt Regency Hotel
Savannah, GA
Pages: 994 - 1000
Cite this article: Dabove, Paolo, Aicardi, Irene, Grasso, Nives, Lingua, Andrea, Ghinamo, Giorgio, Corbi, Cecilia, "Inertial Sensors Strapdown Approach for Hybrid Cameras and MEMS Positioning," Proceedings of IEEE/ION PLANS 2016, Savannah, GA, April 2016, pp. 994-1000.
Full Paper: ION Members/Non-Members: 1 Download Credit
Sign In
Abstract: In this paper the performances (in terms of accuracy) of an hybrid positioning technique, that brings together an image based recognition approach and MEMS (Micro Electro-Mechanical Systems) technology, will be investigated. The image recognition based (IRB) positioning has already been well investigated in the past and represents a good technology for navigation in GNSS denied environment, like indoor or urban canyon, where GNSS accuracy is poor (tens of meters accuracy). However, for practical exploitation of IRB positioning with smartphones the following main problems must be taken into account. The first one is the optimization of the battery, that implies a proper use of the frame rate. A second main issue is represented by latencies due to image processing algorithms and visual search solutions that may require a cloud architecture to manage the size of the database. To overcome the above problems in IRB, reduction of the frame rate and latencies compensation, inertial platform built with MEMS technology may be exploited. Two different approaches are taken into account for IRB positioning: the first one is represented by a real-time solution obtainable through single images; the second one is achieved from a commercial software that locates all the frames (not suitable for real time applications). The paper presents a methodology that fuses IRB with MEMS measurements. Results have shown that with single IRB fixes fused together with inertial navigation the standard errors with the 95th percentile are about 1.66 m, 2.36 m and 3.16 m if the interval between two IRB localizations are 1, 2 and 5 sec respectively, while these errors decrease up to 0.64 m, 1.79 m and 2.05 m if the commercial hybrid absolute and relative orientation solution is considered. We can affirm that the weakness in single photograms positioning approach resides in poor orientation estimation of photograms that can grow up to some tens of degrees when poor geometry is picked in the camera scene. With this study we have demonstrate that the hybrid IRB positioning technique coupled with an INS instrument is useful for indoor navigation because errors are less than 2 m and 1 m, with intervals of about 2 s and 1 s between two images, respectively.