UAV Indoor Positioning based on WiGig Beam Fingerprint

• Main Authors: Hong, Pei-Yuan, 洪培元
• Other Authors: Li, Chi-Yu
• Format: Others
• Language: en_US
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/b8ejhg

碩士 === 國立交通大學 === 資訊科學與工程研究所 === 108 === Positioning system is a key technology to enable Unmanned Aerial Vehicle (UAV) to execute missions autonomously, e.g., landing and aerial photography. UAV can use Global Positioning System (GPS) to know its outdoor location, but the satellite signals are blocked in indoor environments. Current vision-based solutions perform poorly in bad illumination conditions. Other wireless methods do not cover UAV factors, such as low power and flying behaviors. To this end, we propose an UAV positioning system using commodity low-power WiGig devices. We formulate fingerprints by leveraging the diversity of TX/RX beam combinations and their link signal strengths. We apply weighted k-nearest neighbors into overcoming some outlier fingerprint predictions, and use a particle filter to emulate the UAV motions. We prototype and evaluate the system on our UAV platform. The evaluation result shows that it can achieve 90th percentile positioning error below 1 m with various velocity sensor errors.