Long-term Visual Localization using Semantically Segmented Images

TL;DR

This paper presents a semantic segmentation-based approach for long-term visual localization in autonomous vehicles, achieving comparable accuracy to traditional feature-based methods while reducing map storage requirements.

Contribution

It introduces a novel semantic segmentation-based localization method that does not rely on traditional feature descriptors, enabling efficient and robust long-term localization.

Findings

Semantic maps require less storage space than traditional feature maps.

The method achieves localization errors below 1 meter most of the time.

Performs comparably to SIFT-based methods across seasonal variations.

Abstract

Robust cross-seasonal localization is one of the major challenges in long-term visual navigation of autonomous vehicles. In this paper, we exploit recent advances in semantic segmentation of images, i.e., where each pixel is assigned a label related to the type of object it represents, to attack the problem of long-term visual localization. We show that semantically labeled 3-D point maps of the environment, together with semantically segmented images, can be efficiently used for vehicle localization without the need for detailed feature descriptors (SIFT, SURF, etc.). Thus, instead of depending on hand-crafted feature descriptors, we rely on the training of an image segmenter. The resulting map takes up much less storage space compared to a traditional descriptor based map. A particle filter based semantic localization solution is compared to one based on SIFT-features, and even with large seasonal variations over the year we perform on par with the larger and more descriptive SIFT-features, and are able to localize with an error below 1 m most of the time.