Vision aided motion estimation for unmanned helicopters in GPS denied environments

Determining the motion of an unmanned aerial vehicle in GPS-denied environments is a challenging work. In this paper, we present a systematic design and implementation of a vision aided motion estimation approach for an unmanned helicopter in such a condition. A hierarchical vision scheme is proposed to detect a structured landmark, and find the correspondence between the 3D reference points and the projected 2D image points. Based on the obtained correspondence, a motion estimation scheme is presented to compute the relative position and velocity of the vehicle with respect to the local reference. The robust and accurate estimates are achieved by using the Kalman filter fusing the vision information with outputs of the inertial measurement unit (IMU). The robustness and efficiency of the proposed motion estimation approach is verified by using the data collected in ground and flight tests.