Combining visual natural markers and IMU for improved AR based indoor navigation

摘要

The operation and maintenance phase is the longest and most expensive life-cycle period of buildings and facilities. Operators need to carry out activities to maintain equipment to prevent functionality failures. Although some software tools have already been introduced, research studies have concluded that (1) facility handover data is still predominantly dispersed, unformatted and paper-based and (2) hence operators still spend 50% of their on-site work on target localization and navigation. To improve these procedures, the authors previously presented a natural marker-based Augmented Reality (AR) framework that digitally supports facility maintenance operators when navigating indoors. Although previous results showed the practical potential, this framework fails if no visual marker is available, if identical markers are at multiple locations, and if markers are light emitting signs. To overcome these shortcomings, this paper presents an improved method that combines an Inertial Measurement Unit (IMU) based step counter and visual live video feed for AR based indoor navigation support. In addition, the AR based marker detection procedure is improved by learning camera exposure times in case of light emitting markers. A case study and experimental results in a controlled environment reveal the improvements and advantages of the enhanced framework.