2D SLAM Algorithms Characterization, Calibration, and Comparison Considering Pose Error, Map Accuracy as Well as CPU...

Area of Science:

• Robotics
• Computer Science
• Statistical Analysis

Background:

• Simultaneous Localization and Mapping (SLAM) is crucial for autonomous systems.
• Existing SLAM algorithms lack standardized, statistically robust comparison methods.
• Performance evaluation often relies on qualitative assessments or limited metrics.

Purpose of the Study:

• To propose a novel statistical framework for characterizing, calibrating, and comparing 2D SLAM algorithms.
• To provide a quantitative and statistically validated method for SLAM algorithm assessment.
• To enhance confidence in the comparative analysis of SLAM algorithm performance.

Main Methods:

• Utilized descriptive and inferential statistics for robust evaluation.
• Performed Plackett-Burman and factorial experiments for algorithm characterization.
• Applied hypothesis testing and the central limit theorem for calibration and enhancement.

Main Results:

• Established a statistically sound methodology for SLAM algorithm comparison.
• Quantified performance differences across five leading 2D SLAM algorithms (Cartographer, Gmapping, HECTOR-SLAM, KARTO-SLAM, RTAB-Map).
• Evaluated algorithms based on pose error, map accuracy, CPU, and memory usage.

Conclusions:

• The proposed method offers strong statistical evidence for SLAM algorithm comparison.
• The framework enables reliable assessment of algorithm behavior and resource utilization.
• This work provides a foundation for more rigorous and reproducible SLAM research.