Design of Indoor Real-time Positioning on Embedded Platform

Indoor positioning is becoming a very popular research direction in recent years. In this paper, we realize real-time trajectory calculation through the fusion of inertial and vision. We have two experiments in different environments and we find that the algorithm’s error is below 50% of pure inertial at least. It means that the algorithm mentioned in this paper has a higher accuracy. Besides, we run this algorithm on an embedded platform built by ourselves. This platform is based on raspberry pi, which greatly reduces costs of indoor positioning. In this way, algorithm need not to be ran on the bulky computer.