Visual Measurement-based Autonomous Navigation

Providing Position Location Information (PLI) for individual warfighters increases their effectiveness, enables much greater cooperation between forces, and provides strategic information to commanders. Current technologies which attempt to provide PLI use Inertial Measurement Units (IMU) and Global Positioning System (GPS). This combination is undesirable due to weight, size, and power issues. To address this critical challenge, Intelligent Automation Inc. (IAI), along with Prof. Carlo Tomasi of Duke University and Boeing Research and Technology (BR&T), propose to develop SeeStar, an accurate and efficient visual navigation system (VNS). SeeStar has the following key innovations that will produce accurate and stable results in a practical form factor: • Uses three camera, two forward looking cameras acting as a stereo pair and one off angle camera. In our team''s past work [6], it was shown that a wide field of view and high resolution are both needed for high accuracy. • Recover information from blurred images caused by uneven and extraneous motions instead of discarding them. • Prevent unbounded growth of error by detecting and recognizing previously seen landmarks. • Use provably unbiased motion estimation algorithm.