Visual EKF-SLAM from Heterogeneous Landmarks.

Area of Science:

• Robotics
• Computer Vision
• Artificial Intelligence

Background:

• Accurate robot localization is crucial for many applications using embedded sensor data.
• Simultaneous Localization and Mapping (SLAM) fuses spatial and temporal data for navigation in unknown environments.
• Traditional visual SLAM often relies on point features, potentially limiting accuracy.

Purpose of the Study:

• To develop a comprehensive visual SLAM solution integrating both point and line landmarks.
• To improve the accuracy of camera localization by utilizing heterogeneous landmarks.
• To enhance both the front-end and back-end components of the SLAM system.

Main Methods:

• Implemented a heterogeneous landmark-based Extended Kalman Filter SLAM (EKF-SLAM).
• Managed a combined map of point and line landmarks.
• Developed an integrated front-end active-search process for linear landmarks.
• Evaluated landmark parametrization and heterogeneity's impact on localization accuracy.

Main Results:

• The proposed method demonstrated improved accuracy in camera localization.
• The integration of line landmarks enhanced the overall SLAM performance.
• The heterogeneous landmark approach proved effective in complex environments.

Conclusions:

• Combining point and line landmarks in visual SLAM offers significant advantages for localization accuracy.
• The developed system provides a robust and accurate solution for robot navigation.
• This approach advances the field of visual SLAM by incorporating richer environmental features.