A Functional Estimation Testbed for Robust Vision-Aided Navigation

Many defense platforms depend on GPS for precise navigation and targeting. Due to proliferation of inexpensive GPS jamming devices and emergence of anti-satellite capabilities, there is increased demand for sensor fusion algorithms which provide GPS-quality localization and are robust to GPS jamming, cutouts, and hardware failure. Our Phase I effort analyzed GPS/INS coupling architectures, and determined that feedback from a multi-sensor optimization algorithm can increase efficiency and accuracy of tracking visual feature locations or other variables related to a platform trajectory. This observation has resulted in a new functional estimation framework for vision-aided navigation. In Phase II, we will develop a testbed for design and evaluation of multiple algorithms in combination with different sensors and aircraft platforms. Our performance metrics will enable benchmarking against both simulated data and captive flight data. This testbed may be used by the Air Force and industry for level-playing field testing of different sensors and navigation algorithms against a common standard of accuracy and effectiveness.  SSCI will partner with Dr. Mubarak Shah and Dr. Niels da Vitoria Lobo at the University of Central Florida. To encourage future involvement of the vision-aided navigation research community and industry, it is our intention to publish this work in an academic conference or journal. BENEFITS: Development of tightly coupled image processing algorithms will greatly enhance robustness and precision of guidance and navigation systems, especially in GPS-denied environments. In addition, development of a testbed for a large class of platforms, sensors, and algorithms will make it possible to swap components and evaluate performance for each combination. This work will engage academia and industry and foster new algorithms to be transitioned into flight systems by providing standards for subsystem interfacing and for direct performance comparison.