Our research goal is to implement an indoor positioning system with 1 or 3 meters accuracy for the pedestrian with smartphone. Our proposed system must be practical, easy and accurate. Considering those requirements, our proposed solution is to integrate a new vision based indoor positioning technology without image DB and an advanced PDR technology with motion recognition. To verify proposed technology, we implemented the system using android smartphone and had several tests. In this paper, we propose a vision based positioning technology without image DB(database) for the initial position of PDR. Generally, vision based positioning technology is to compare user snapshot image with geotagged images from DB. However our technology is to compare user snapshot image with indoor map only. Our proposed vision based positioning technology is to compare features from images around pedestrian with indoor map. In the real hallway tests, the accuracy of our estimated position is less than 2m. In this paper, we present a 3D pedestrian dead reckoning (PDR) system for indoor navigation using a smartphone. The motion-free PDR conduct a pedestrian tracking regardless of any motion. We define 6 pedestrian motions and each motion is recognized using artificial neural network (ANN) classifier. A proper PDR algorithm is operated according to the recognized motion. A height of pedestrian is estimated using barometric pressure sensor in the smartphone. To demonstrate the performance of proposed system, we conduct a field test and its results suggest that this system could be utilized for the 3D indoor navigation system. By integrating a vision technology and a motion recognition based PDR technology, we could implement a seamless pedestrian positioning system for any pedestrian with a smartphone. We implemented our technology into the android based smartphone. A pedestrian got his initial position using vision based positioning technology without image DB. And he tracked his position using advanced PDR technology with motion recognition. In our real tests, the positioning accuracy was about several meters and it gave seamless positioning solution in the indoor environments.
展开▼
