The key towards achieving real-time performance is the derivation of a novel measurement model, which expresses these constraints without including the 3D feature position in the filter state vector. The resulting EKF-based estimation algorithm is capable of high-precision, real-time pose estimation, and thus can employed in demanding applications such as UAV localization, vision-aided navigation for precise spacecraft landing, tracking the pose of mobile robots, etc.
This is pretty cool.
No comments:
Post a Comment