详情

原文传递 COMBINED STEREO VISION AND INERTIAL NAVIGATION FOR AUTOMATED AERIAL REFUELING.

题名：	COMBINED STEREO VISION AND INERTIAL NAVIGATION FOR AUTOMATED AERIAL REFUELING.
作者：	Johnson, D. T.
关键词：	Aerial refueling, Navigation
摘要：	The lack of aerial refueling is a fundamental shortcoming of current Unmanned Aerial Vehicles (UAVs). UAV deployment continues to increase but the corresponding operational capabilities are limited by an inability to conduct extended operations. To alleviate this limitation and achieve the Air Forces global mobility and global strike missions, UAV must procure the capability to perform Automated Aerial Refueling (AAR). Such capability requires computing the precise relative attitude and separation (pose) between the tanker and unmanned receiver. This relative pose is fundamentally required to hold the receiver within the tankers refueling envelope throughout the duration of the refueling procedure. The Global Positioning System (GPS) is one viable data source from which this relative navigation solution may be computed; however, GPS signals may be degraded or denied within an operational environment. Other sensors, such as Light Detection and Ranging (LiDAR), Machine Vision (MV), and Stereo Machine Vision (SMV), are also viable candi- dates. Other work has utilized dierent combinations of these sensors utilizing a Kalman Filter (KF) to compensate for individual weaknesses. This work presents an integrated SMV / Inertial Navigation System (INS) as a method to compute the relative pose solution; optionally, GPS may be also integrated, if it is available. This integration is accomplished through design of an Extended Kalman Filter (EKF) used to bolster the SMV via INS/GPS. SMV errors caused by noise and processing are reduced by the EKF and the SMV prevents the INS from drifting. The INS/GPS/SMV EKF solution was generated and tested in a virtual simulation employing a stereo camera pair observing an approaching C12 aircraft in a custom 3D Virtual World (3DVW). Experimental results demonstrate that the EKF is able to combine SMV and GPS with INS individually or in tandem to achieve Root Mean Squared Error (RMSE) less than 2 cm when both sensors are available. When GPS is denied or lost the EKF with SMV achieves RMSE less than 10 cm at distances greater than 20 m from the stereo camera system.
报告类型：	科技报告