Multi-Objective Optimization of Loop Closure Detection Parameters for Indoor 2D Simultaneous Localization and Mapping

Area of Science:

• Robotics
• Computer Vision
• Artificial Intelligence

Background:

• Tuning loop closure detection parameters is critical for accurate 2D graph-based Simultaneous Localization and Mapping (SLAM) in indoor environments.
• Existing methods often lack efficient parameter tuning strategies, leading to suboptimal map quality.
• Accurate mapping is essential for reliable robot navigation and task execution.

Purpose of the Study:

• To propose a multi-objective optimization method for tuning loop closure detection parameters in 2D graph-based SLAM.
• To enhance the quality of maps generated by SLAM algorithms, particularly in scenarios with nested loops.
• To provide a robust evaluation framework for map quality without requiring ground truth data.

Main Methods:

• Integration of the Karto SLAM algorithm with a multi-objective optimization algorithm.
• Development of three quantitative map quality metrics: proportion of occupied grids, number of corners, and amount of enclosed areas.
• Validation of the proposed method using four datasets and two real-world environments.

Main Results:

• The proposed multi-objective optimization method successfully improved map quality in all tested scenarios.
• The evaluation metrics effectively identified mapping errors like overlaps, blurring, and misalignment.
• The method demonstrated robustness in improving map accuracy for indoor 2D graph-based SLAM.

Conclusions:

• The proposed method offers an effective solution for automatically tuning loop closure detection parameters in 2D SLAM.
• The developed evaluation metrics provide a valuable tool for assessing map quality without ground truth.
• The approach has potential for broader applications in optimizing other SLAM parameters to enhance map generation.