Self-Supervised Robust Scene Flow Estimation via the Alignment of Probability Density Functions
This addresses robust 3D motion estimation in point clouds for applications like autonomous driving, offering a novel self-supervised approach that improves accuracy in real-world environments.
The paper tackles scene flow estimation for consecutive point clouds by converting discrete points into continuous probability density functions using Gaussian mixture models and aligning them via Cauchy-Schwarz divergence, achieving state-of-the-art performance among self-supervised methods on FlyingThings3D and KITTI datasets.
In this paper, we present a new self-supervised scene flow estimation approach for a pair of consecutive point clouds. The key idea of our approach is to represent discrete point clouds as continuous probability density functions using Gaussian mixture models. Scene flow estimation is therefore converted into the problem of recovering motion from the alignment of probability density functions, which we achieve using a closed-form expression of the classic Cauchy-Schwarz divergence. Unlike existing nearest-neighbor-based approaches that use hard pairwise correspondences, our proposed approach establishes soft and implicit point correspondences between point clouds and generates more robust and accurate scene flow in the presence of missing correspondences and outliers. Comprehensive experiments show that our method makes noticeable gains over the Chamfer Distance and the Earth Mover's Distance in real-world environments and achieves state-of-the-art performance among self-supervised learning methods on FlyingThings3D and KITTI, even outperforming some supervised methods with ground truth annotations.