Vision-IMU based collaborative control of a blind UAV

VI-RPE: Visual-Inertial Relative Pose Estimation for Aerial Vehicles

“…Compared to the traditional multinomial approach, this method results to more uniformly distributed particles and smaller variance of the particles' distribution, which improves the quality of the resampling. Similarly to the initialization, the final pose for the UAV is obtained via optimization using Equation (8). Here the initial pose guess for the optimization is set to the pose with more particles after the resampling.…”

“…Moreover, the high accuracy of these tracking systems make this the most popular approach to accurate pose tracking in indoor confined spaces. April-tag-like [7] planar markers are also commonly used, however, mounting planar markers onboard a UAV, severally influence its dynamics and only very small tags can be placed on UAVs, as in [8], leading to inaccuracies and a small operational range. In practice, planar markers are only used on non-aerial vehicles or static objects, such as walls or on the ground.…”