| Popis: |
Honeywell’s Vision-Aided Navigation system is built on MAURO (Map Abstracted Universal Representation of Orthoimagery), an internally developed image correlation algorithm, for airborne vision navigation in GPS-denied environments. The system generates absolute position measurements from a downward looking camera by correlating the camera image to a compressed map. Previously demonstrated under DARPA’s SECTR program, Honeywell continues to invest in the technology and recently demonstrated airborne GPS-denied navigation in real-time at PNTAX20. Honeywell’s approach offers several key advantages, including, but not limited to, sensor agnostic operation and very low map storage. The low map footprint provides ability to store whole-theater, a priori, geo-referenced map representations on any platform, eliminating the need for mission pre-planning and database customization for GPS-denied navigation. Sensor agnostic operation supports cross-modal correlation using visual band, infrared, or other imaging sensors. The same map can be used for navigation with different sensor types, supporting platform/mission versatility and future expansion to other sensor modalities. This approach is also very efficient, using operations that are highly optimized for GPU, reducing the processing time on affordable embedded hardware while increasing the robustness and accuracy. Honeywell has already demonstrated that this approach is robust to environmental and lighting changes, as well as map staleness. Honeywell’s approach, unlike SLAM or visual odometry, does not drift over time, but provides an absolute position fix. Honeywell’s Vision-Aided navigation system can be combined with other alternative navigation technologies to provide integrity, redundancy, and full mission coverage for all GPS-Denied conditions. These alternative navigation technologies include celestial vision, magnetic correlation, and novel visual odometry methods. During PNTAX20, Vision-Aided navigation system was demonstrated using the Honeywell Embraer E170 in four simultaneous configurations. Two different, low cost LWIR sensors were separately used with a tactical MEMS IMU and a navigation grade RLG IMU. The a priori, geo-referenced map database was generated from visible spectrum satellite imagery. Real-time and playback results demonstrating the GPS-denied performance of the Vision-aided system using the Teledyne DALSA Calibir™ and FLIR Boson® cameras will be presented. Summary results of over 50 hours of Honeywell collected data will also be presented, demonstrating performance across many terrain types and altitudes. |
