IR Algorithms for Video-Aided Navigation

This project will develop a FLIR-based visual odometry capability for Unmanned Air Vehicles, giving them a robust autonomous navigation capability. The system will allow UAVs to maintain an accurate estimate of their own motion in the absence of GPS, with INS drift, and in day or night operating conditions. Key aspects of our development strategy are to (1) fully exploit all available FLIR phenomenology to make the system most generally applicable, (2) consider a wide variety of navigation system sensory inputs to be useful to many UAV platforms, and (3) not be tied to a single motion estimation approach but rather evaluate several candidate approaches in the development of our preferred approach for real-time implementation and demonstration on Phase II. During Phase I we will develop feature extraction methods appropriate for FLIR and motion estimation, conduct trade studies of motion estimation approaches using FLIR features, and develop a preferred approach for Phase II.