Point-cloud-based place recognition using CNN feature extraction

TL;DR

This paper introduces a deep learning-based point-cloud place recognition system that leverages pre-trained CNN features from range images, achieving robustness and invariance, along with a new comprehensive dataset for evaluation.

Contribution

It presents a novel approach using pre-trained CNNs for point-cloud feature extraction without fine-tuning, and provides a new dataset for benchmarking place recognition methods.

Findings

Significant improvement over hand-crafted features.

System is illumination and rotation invariant.

Robust against unrelated moving objects.

Abstract

This paper proposes a novel point-cloud-based place recognition system that adopts a deep learning approach for feature extraction. By using a convolutional neural network pre-trained on color images to extract features from a range image without fine-tuning on extra range images, significant improvement has been observed when compared to using hand-crafted features. The resulting system is illumination invariant, rotation invariant and robust against moving objects that are unrelated to the place identity. Apart from the system itself, we also bring to the community a new place recognition dataset containing both point cloud and grayscale images covering a full $360^\circ$ environmental view. In addition, the dataset is organized in such a way that it facilitates experimental validation with respect to rotation invariance or robustness against unrelated moving objects separately.