UAS pose estimation based on image-assisted total stations

Abstract

The estimation of the pose, which is the position and orientation of a multi-sensor system, is a crucial part of the mapping process. The pose of an Uncrewed Aerial Systems (UAS) is typically estimated using satellite and inertial navigation systems, but it can also be estimated based on image-assisted total station (IATS) observations. Independent and absolute pose estimation based on IATS complements conventional GNSS/IMU-based orientation estimates and can improve the orientation. This is of particular relevance for survey scenarios where neither absolute nor relative ground control is available, which applies to most bathymetric survey scenes. In our study, we introduce a methodology for IATS-based orientation estimation alongside a simulation study and a practical implementation. We leverage direction vectors, derived from image observations of the IATS, to spherical targets on the UAS. We introduce these vectors in a least-squares adjustment to estimate the orientation angles of the platform. In a static, terrestrial experiment, we reached standard deviations of roll, pitch, and yaw angles between 0.10 deg to 0.15 deg , which confirms the uncertainty estimates of our Monte-Carlo simulation. For a real-world airborne experiment at the pre-alpine Pielach River, where a multi-copter UAV was used, we obtained deviations of 0.26 deg to 0.49 deg between two independent IATS. The results show that the most accurate angle is the yaw, with a deviation of 0.26 deg , thus, validating the complementarity of the proposed method to the IMU. When evaluating the IATS-based orientations with poses estimated in a bundle-block adjustment, the deviations range from 0.5 deg to 1.0 deg.