PointFlowNet: Learning Representations for Rigid Motion Estimation from Point Clouds

TL;DR

PointFlowNet is a deep learning model that estimates 3D scene flow and object motion directly from unstructured point clouds, addressing challenges of rigid motion inference with a novel translation equivariant representation.

Contribution

The paper introduces a translation equivariant representation for rigid motion estimation from point clouds and a large augmented dataset for training deep networks in this domain.

Findings

Outperforms classic and learning-based methods in robustness

Effectively estimates 3D scene flow and object motion from point clouds

Handles occlusions and sensor noise realistically

Abstract

Despite significant progress in image-based 3D scene flow estimation, the performance of such approaches has not yet reached the fidelity required by many applications. Simultaneously, these applications are often not restricted to image-based estimation: laser scanners provide a popular alternative to traditional cameras, for example in the context of self-driving cars, as they directly yield a 3D point cloud. In this paper, we propose to estimate 3D motion from such unstructured point clouds using a deep neural network. In a single forward pass, our model jointly predicts 3D scene flow as well as the 3D bounding box and rigid body motion of objects in the scene. While the prospect of estimating 3D scene flow from unstructured point clouds is promising, it is also a challenging task. We show that the traditional global representation of rigid body motion prohibits inference by CNNs, and propose a translation equivariant representation to circumvent this problem. For training our deep network, a large dataset is required. Because of this, we augment real scans from KITTI with virtual objects, realistically modeling occlusions and simulating sensor noise. A thorough comparison with classic and learning-based techniques highlights the robustness of the proposed approach.