Deep Bayesian ICP Covariance Estimation
This work addresses a specific issue in robotics and computer vision for improving sensor fusion accuracy, though it is incremental as it builds on existing deep learning methods for point clouds.
The paper tackles the problem of covariance estimation for the Iterative Closest Point (ICP) algorithm in point cloud registration, which is crucial for state estimation and sensor fusion, by proposing a deep Bayesian approach that learns an error model from data to estimate heteroscedastic aleatoric and epistemic uncertainties, achieving good results in LiDAR odometry evaluations on various datasets.
Covariance estimation for the Iterative Closest Point (ICP) point cloud registration algorithm is essential for state estimation and sensor fusion purposes. We argue that a major source of error for ICP is in the input data itself, from the sensor noise to the scene geometry. Benefiting from recent developments in deep learning for point clouds, we propose a data-driven approach to learn an error model for ICP. We estimate covariances modeling data-dependent heteroscedastic aleatoric uncertainty, and epistemic uncertainty using a variational Bayesian approach. The system evaluation is performed on LiDAR odometry on different datasets, highlighting good results in comparison to the state of the art.