Submodular Optimization for Keyframe Selection & Usage in SLAM

2024-10-08

arXiv.org

AI Summary

Traditional SLAM systems rely on inefficient heuristic rules for keyframe selection, leading to resource waste and suboptimal performance. This paper proposes the first online submodular optimization framework tailored for SLAM keyframe decision-making. Our approach features: (1) a decoupled dual-objective mechanism that separately optimizes localization accuracy and mapping fidelity; (2) a constraint-aware subgraph reconfiguration method enabling high-fidelity map summarization under strict memory budgets; and (3) end-to-end integration into LiDAR-SLAM, supporting incremental keyframe evaluation and dynamic subgraph updates. Experiments demonstrate that our method reduces keyframe count by over 40% on average while preserving localization accuracy—achieving zero degradation in pose estimation—yet significantly improving per-frame processing efficiency. Moreover, it generates compact, high-quality environmental summaries even under stringent size constraints.

Technology Category

• Search and Optimization: Learning to Search
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Keyframes are LiDAR scans saved for future reference in Simultaneous Localization And Mapping (SLAM), but despite their central importance most algorithms leave choices of which scans to save and how to use them to wasteful heuristics. This work proposes two novel keyframe selection strategies for localization and map summarization, as well as a novel approach to submap generation which selects keyframes that best constrain localization. Our results show that online keyframe selection and submap generation reduce the number of saved keyframes and improve per scan computation time without compromising localization performance. We also present a map summarization feature for quickly capturing environments under strict map size constraints.

Problem

• Optimizes keyframe selection in SLAM to reduce waste
• Improves submap generation for better localization constraints
• Enables efficient map summarization under size limits

Innovation

• Novel keyframe selection strategies
• Submap generation optimizing localization
• Efficient map summarization under constraints