マルチコプタを利用した橋梁点検システムの開発

―複数センサ情報を用いた3次元位置推定技術の開発―

抄録

We are developing a multicopter system for bridge inspection to take high-resolution images with a mounted camera while flying along a wall at a short distance. In the case of a 50m class bridge pier, it may be quite difficult for the user to remotely operate the multicopter so as not to hit the wall because of the long distance from the multicopter. In this paper we discuss the assist control system that keeps the distance between the wall and the multicopter automatically. Since GPS cannot be used under the bridge, the ultrasonic sensor was alternatively used to measure the distance. However, if the airframe is drifted by strong wind, it would collide with the wall due to the low sampling rate of the ultrasonic sensor. We thus propose a method to improve the stability of proximity control by using multiple sensors such as laser range finder and barometer with high sampling rate. We show the effectiveness of the proposed method by simulation using MAVROS.