LFM: A Lightweight LCD Algorithm Based on Feature Matching between Similar Key Frames.

Area of Science:

• Robotics
• Computer Vision
• Artificial Intelligence

Background:

• Simultaneous Localization and Mapping (SLAM) systems require accurate Loop Closure Detection (LCD) to correct accumulated errors.
• Traditional LCD methods often struggle with accuracy and efficiency in complex environments.

Purpose of the Study:

• To develop a highly accurate and efficient Loop Closure Detection (LCD) algorithm for robotic SLAM using deep learning.
• To improve feature matching accuracy between similar images for robust loop closure identification.

Main Methods:

• A novel lightweight Convolutional Neural Network (CNN) was designed for key frame target detection.
• Key frames were binary classified using deep learning.
• An improved lightweight feature matching network based on Transformer was employed for loop closure judgment.

Main Results:

• The proposed LFM-LCD algorithm demonstrated superior accuracy and recall rates compared to traditional methods in indoor SLAM scenarios.
• The method achieved these improvements while maintaining a comparable number of parameters and computational cost.
• Experimental results validate the effectiveness of the deep learning-based approach for LCD.

Conclusions:

• The developed LFM-LCD algorithm offers a significant advancement in robotic SLAM accuracy.
• This research provides a new, promising direction for Loop Closure Detection in SLAM systems.
• Further improvements are anticipated with ongoing advancements in deep learning technologies.