Learning the Next Best View for 3D Point Clouds via Topological Features

TL;DR

This paper presents a reinforcement learning method using topological features to determine the next best view for capturing detailed 3D point clouds, improving robotic sensor placement.

Contribution

It introduces a novel topology-based information gain metric and demonstrates its effectiveness in guiding sensor views for detailed 3D scanning.

Findings

Enhanced view planning for 3D point clouds

Improved detail capture in noisy sensor data

Public release of datasets and tools

Abstract

In this paper, we introduce a reinforcement learning approach utilizing a novel topology-based information gain metric for directing the next best view of a noisy 3D sensor. The metric combines the disjoint sections of an observed surface to focus on high-detail features such as holes and concave sections. Experimental results show that our approach can aid in establishing the placement of a robotic sensor to optimize the information provided by its streaming point cloud data. Furthermore, a labeled dataset of 3D objects, a CAD design for a custom robotic manipulator, and software for the transformation, union, and registration of point clouds has been publicly released to the research community.