Accurate Mobile Urban Mapping via Digital Map-Based SLAM.

Area of Science:

• Robotics
• Computer Vision
• Geomatics Engineering

Background:

• Conventional high-accuracy mapping relies on expensive sensors or static systems.
• Mobile mapping systems often struggle with Global Positioning System (GPS) limitations.
• Urban environments present challenges due to sporadic GPS availability and complex structures.

Purpose of the Study:

• To develop an accurate urban map generation system using digital map-based Simultaneous Localization and Mapping (SLAM).
• To achieve sub-meter accuracy for 3D and lane maps using a mobile platform.
• To integrate complementary data sources (GPS, digital maps, sensors) for robust localization and mapping.

Main Methods:

• Leveraged SLAM by modeling GPS and digital map data as complementary measurements.
• Developed an efficient graph SLAM algorithm for real-time performance.
• Implemented a motion-adaptive Inverse Perspective Mapping (IPM) model integrated with SLAM.

Main Results:

• Achieved a real-time mobile mapping framework targeting sub-meter accuracy.
• Demonstrated stable generation of bird's-eye view images despite significant motion.
• Validated the framework through long-distance urban testing and Real-Time Kinematic (RTK)-GPS evaluation.

Conclusions:

• The proposed SLAM-based approach effectively generates accurate urban maps in real-time.
• Integrating sporadic GPS and digital map data enhances mobile mapping accuracy.
• The system provides a viable solution for high-accuracy mobile mapping in challenging urban environments.