Road Mapping and Localization using Sparse Semantic Visual Features

TL;DR

This paper introduces a novel visual mapping and localization method for autonomous vehicles that leverages deep learning to detect semantic road features, improving pose accuracy and map compactness.

Contribution

It proposes a new approach that uses deep models to detect and model semantic road elements, replacing traditional point features for enhanced accuracy and efficiency.

Findings

Outperforms traditional feature-based methods in accuracy

Effective in diverse public datasets and real testing environments

Provides a complete pipeline integrating detection, modeling, and optimization

Abstract

We present a novel method for visual mapping and localization for autonomous vehicles, by extracting, modeling, and optimizing semantic road elements. Specifically, our method integrates cascaded deep models to detect standardized road elements instead of traditional point features, to seek for improved pose accuracy and map representation compactness. To utilize the structural features, we model road lights and signs by their representative deep keypoints for skeleton and boundary, and parameterize lanes via piecewise cubic splines. Based on the road semantic features, we build a complete pipeline for mapping and localization, which includes a) image processing front-end, b) sensor fusion strategies, and c) optimization backend. Experiments on public datasets and our testing platform have demonstrated the effectiveness and advantages of our method by outperforming traditional approaches.