Evaluating geometric accuracy of NeRF reconstructions compared to SLAM method

TL;DR

This paper evaluates the geometric accuracy of NeRF-based scene reconstructions against traditional SLAM methods, focusing on their ability to accurately measure a PVC cylinder using different data sources.

Contribution

It provides a comparative analysis of NeRF reconstructions trained on different datasets versus SLAM in terms of noise and metric accuracy.

Findings

NeRF can achieve comparable geometric accuracy to SLAM.

NeRF trained on iPhone data performs well in scene reconstruction.

SLAM shows lower noise in metric measurements.

Abstract

As Neural Radiance Field (NeRF) implementations become faster, more efficient and accurate, their applicability to real world mapping tasks becomes more accessible. Traditionally, 3D mapping, or scene reconstruction, has relied on expensive LiDAR sensing. Photogrammetry can perform image-based 3D reconstruction but is computationally expensive and requires extremely dense image representation to recover complex geometry and photorealism. NeRFs perform 3D scene reconstruction by training a neural network on sparse image and pose data, achieving superior results to photogrammetry with less input data. This paper presents an evaluation of two NeRF scene reconstructions for the purpose of estimating the diameter of a vertical PVC cylinder. One of these are trained on commodity iPhone data and the other is trained on robot-sourced imagery and poses. This neural-geometry is compared to state-of-the-art lidar-inertial SLAM in terms of scene noise and metric-accuracy.