Analysis of different feature selection criteria based on a covariance convergence perspective for a SLAM algorithm

Area of Science:

• Robotics
• Computer Vision
• Artificial Intelligence

Background:

• Simultaneous Localization and Mapping (SLAM) is crucial for mobile robot navigation.
• Existing SLAM algorithms can be computationally intensive, impacting real-time performance.
• Feature selection in SLAM is often arbitrary, affecting algorithm convergence and accuracy.

Purpose of the Study:

• To introduce non-arbitrary feature selection techniques for SLAM.
• To enhance SLAM convergence using significant features.
• To reduce the computational load and processing time of SLAM algorithms.

Main Methods:

• Implemented a sequential Extended Kalman Filter (EKF) SLAM algorithm.
• Developed feature selection criteria based on SLAM convergence significance.
• Applied feature selection during the correction stage of the EKF-SLAM algorithm.

Main Results:

• Demonstrated a reduction in SLAM processing time through restricted feature correction.
• Evaluated the performance of different feature selection techniques via experiments.
• Showcased map reconstruction capabilities of the proposed methods in an outdoor environment.

Conclusions:

• The proposed non-arbitrary feature selection significantly improves SLAM efficiency.
• The techniques enhance convergence and reduce computational demands for mobile robot mapping.
• The methods are effective in diverse outdoor environments and not limited to specific feature types.