sSLAM: Speeded-Up Visual SLAM Mixing Artificial Markers and Temporary Keypoints.

Area of Science:

• Robotics
• Computer Vision
• Artificial Intelligence

Background:

• Visual SLAM (vSLAM) typically uses environmental keypoints as landmarks.
• Keypoints are often unstable due to environmental changes like shadows and moving objects.
• Existing methods combine keypoints with artificial markers but retain unnecessary keypoints, increasing computational load.

Purpose of the Study:

• To propose a novel vSLAM approach that efficiently combines keypoints and artificial markers.
• To reduce computing time and memory requirements without sacrificing tracking accuracy.
• To develop a faster vSLAM system compared to state-of-the-art methods.

Main Methods:

• The proposed system initially maps environments using both keypoints and artificial markers.
• Keypoints are discarded after map creation, retaining only long-lasting artificial markers.
• Temporary keypoints are generated during tracking for localization alongside marker information.

Main Results:

• The novel vSLAM approach significantly reduces computing time and memory usage.
• Tracking accuracy is maintained without noticeable degradation.
• Experimental results show favorable speed comparisons against ORB-SLAM2, ORB-SLAM3, OpenVSLAM, and UcoSLAM.

Conclusions:

• The developed vSLAM method offers a substantial improvement in efficiency by utilizing a map of only long-lasting features (markers).
• The system achieves faster performance than existing vSLAM approaches while maintaining comparable accuracy.
• This approach presents a viable solution for efficient and accurate real-time localization and mapping.