CELLO-3D: Estimating the Covariance of ICP in the Real World
This work addresses uncertainty estimation for ICP in robotics and computer vision, but it is incremental as it improves upon existing covariance estimation methods.
The paper tackles the problem of accurately estimating the covariance of Iterative Closest Point (ICP) registrations for uncertainty in state estimation frameworks, using a data-driven approach with over 5.1 million registrations on real 3D point clouds, and shows that it outperforms existing closed-form solutions and provides predictions consistent with observed trajectories.
The fusion of Iterative Closest Point (ICP) reg- istrations in existing state estimation frameworks relies on an accurate estimation of their uncertainty. In this paper, we study the estimation of this uncertainty in the form of a covariance. First, we scrutinize the limitations of existing closed-form covariance estimation algorithms over 3D datasets. Then, we set out to estimate the covariance of ICP registrations through a data-driven approach, with over 5 100 000 registrations on 1020 pairs from real 3D point clouds. We assess our solution upon a wide spectrum of environments, ranging from structured to unstructured and indoor to outdoor. The capacity of our algorithm to predict covariances is accurately assessed, as well as the usefulness of these estimations for uncertainty estimation over trajectories. The proposed method estimates covariances better than existing closed-form solutions, and makes predictions that are consistent with observed trajectories.