A Flying Robot Localization Method Based on Multi-Sensor Fusion

Abstract

This paper proposes a novel localization method for a power-tower-inspection flying robot based on fusion of vision, IMU and GPS. First, the research background is introduced in relation to a visual localization algorithm derived from 3D-model-based tracking and a coordinate transformation model for related coordinate frames. Then, a multi-sensor fusion-based localization method is presented, in which two collaborative Kalman filters are designed to fuse IMU/GPS and visual information. Finally, experimental results are presented to show the robustness and precision of the proposed method.