SparseLoc: Sparse Open-Set Landmark-based Global Localization for Autonomous Navigation

2025-03-30

AI Summary

To address the high storage and computational overhead of dense LiDAR maps and the poor robustness of sparse methods in GPS-denied global localization, this paper proposes SparseLoc. Our method introduces the first zero-shot sparse semantic topological map generation framework leveraging vision-language foundation models (e.g., CLIP/ViTL), eliminating reliance on manual annotation and dense point clouds. We further design a delay-optimized mechanism to enhance Monte Carlo Localization (MCL) robustness in dynamic and texture-deprived environments. Evaluated on the KITTI dataset, SparseLoc achieves sub-5 m average position error and sub-2° orientation error using only 0.2% of the original point cloud density. It outperforms existing sparse approaches by over 5× in accuracy, matching the precision of dense-map-based methods while drastically reducing memory footprint and computational cost.

Technology Category

• Intelligent Robots: Localization, Mapping, and Navigation
• Planning, Routing, and Scheduling: Planning with Language Models
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Global localization is a critical problem in autonomous navigation, enabling precise positioning without reliance on GPS. Modern global localization techniques often depend on dense LiDAR maps, which, while precise, require extensive storage and computational resources. Recent approaches have explored alternative methods, such as sparse maps and learned features, but they suffer from poor robustness and generalization. We propose SparseLoc, a global localization framework that leverages vision-language foundation models to generate sparse, semantic-topometric maps in a zero-shot manner. It combines this map representation with a Monte Carlo localization scheme enhanced by a novel late optimization strategy, ensuring improved pose estimation. By constructing compact yet highly discriminative maps and refining localization through a carefully designed optimization schedule, SparseLoc overcomes the limitations of existing techniques, offering a more efficient and robust solution for global localization. Our system achieves over a 5X improvement in localization accuracy compared to existing sparse mapping techniques. Despite utilizing only 1/500th of the points of dense mapping methods, it achieves comparable performance, maintaining an average global localization error below 5m and 2 degrees on KITTI sequences.

Problem

• Enables GPS-free precise positioning for autonomous navigation
• Reduces storage and computation needs of dense LiDAR maps
• Improves robustness and generalization in sparse map localization

Innovation

• Uses vision-language models for sparse maps
• Enhances Monte Carlo localization with optimization
• Achieves high accuracy with minimal points