A Simultaneous Localization and Mapping System Using the Iterative Error State Kalman Filter Judgment Algorithm for...

Area of Science:

• Robotics and Autonomous Systems
• Sensor Fusion
• Simultaneous Localization and Mapping (SLAM)

Background:

• Outdoor autonomous mobile robots depend on GPS for localization, but urban environments and signal loss cause significant deviations.
• Existing SLAM systems struggle with GPS inaccuracies and signal interruptions, limiting reliable outdoor navigation.
• Accurate localization is critical for safe and efficient autonomous robot operation in diverse environments.

Purpose of the Study:

• To develop an improved SLAM system addressing GPS inaccuracies and signal loss for outdoor autonomous robots.
• To enhance localization accuracy and robustness in challenging environments with unreliable GPS data.
• To provide a reliable navigation solution for autonomous mobile robots operating in urban and GPS-denied areas.

Main Methods:

• Proposed a SLAM system combining Iterated Extended Kalman Filter (IESKF) and factor graph optimization.
• Utilized LiDAR and Inertial Measurement Unit (IMU) data for front-end pose estimation via IESKF.
• Implemented a GPS signal filtering method based on state and confidence, and integrated processed GPS, IMU, and loop closure factors into a back-end factor graph.

Main Results:

• The proposed SLAM system demonstrated superior localization accuracy compared to FAST-LIO and LIO-SAM on the KITTI dataset and a custom platform.
• Significant performance improvements were observed, particularly in scenarios with intermittent or lost GPS signals.
• The integration of filtered GPS data and sensor fusion enhanced the overall robustness of the localization system.

Conclusions:

• The developed SLAM system effectively overcomes GPS limitations, offering enhanced localization accuracy and reliability for outdoor autonomous robots.
• The proposed sensor fusion and GPS filtering approach provides a robust solution for navigation in challenging environments.
• This work contributes to advancing autonomous navigation capabilities in real-world applications where GPS is often unreliable.