Self-calibrated Navigation System Based on Fusion of Distance Measurement and Visual Odometry

In this paper, a self-calibrated navigation system based on fusion of distance measurement and visual odometry (VO) is developed. The navigation system can locate the agent with low cumulative error in the GNSS-denied environment by using some randomly placed anchors. There are two stages in the system, calibrating and positioning. In the calibrating stage, the location of each anchor is estimated optimally. Meanwhile, a criterion is proposed. It can help judge whether current measurements are sufficient to estimate of the location of each anchor. In the positioning stage, the location of the agent is estimated based on fusing of VO and range measurements between the agent and anchors. The navigation system is validated via data sets and real-world experiments. The results show that the developed system can simultaneously estimate the location of each anchor and the drone accurately and stably.