Stereo visual navigation based on network adjustment of relative orientation parameters

Visual navigation (VN) is an algorithm of estimating the motion of consecutive frames. The motion can be described by the exterior orientation parameters (EOPs) including the attitude and position. However, using a single camera for VN could not recover the real scale. The distance between adjacent images is unknown. Besides, the error would exist in the EOPs due to the error matching. To overcome the above issues, stereo VN is adopted in this research. Stereo VN can recover the real scale according to the pre-calibrated relative orientation of two cameras. However, the error still exists in the solved orientation including the rotation, translation, and scale. Therefore, the network adjustment model based on relative orientation parameters (ROPs) is proposed to optimize the motion of consecutive frames. The proposed network adjustment model is verified by not only simulation data but also the real experiment. Consequently, ROPs are adopted as observations in the model that would update the states of image sequence furthermore. Not only the scale issue could be solved, but also the rotation and translation solution could be improved significantly during the whole operation. In this research, it is worth mentioning that the coordinates of object points are not necessary to be calculated, which means the conventional bundle adjustment is not adopted, but more accurate ROPs still have improved automatically during the process.