OL-SLAM: A Robust and Versatile System of Object Localization and SLAM

Area of Science:

• Robotics and Autonomous Systems
• Sensor Fusion
• Computer Vision

Background:

• Accurate real-time localization and mapping are crucial for autonomous systems.
• Integrating multiple sensor modalities (LiDAR, camera, IMU, GPS) presents challenges in achieving robust performance.
• Existing systems often struggle with precise object localization in dynamic environments.

Purpose of the Study:

• To develop a versatile, real-time system for simultaneous localization and mapping (SLAM) and object localization.
• To fuse data from LiDAR, camera, Inertial Measurement Unit (IMU), and Global Positioning System (GPS) for enhanced accuracy.
• To enable autonomous systems to map unknown environments and track objects with global positioning.

Main Methods:

• A multi-part SLAM subsystem combining LiDAR-inertial, visual-inertial, and GPS-inertial odometry.
• Global pose graph optimization for refining the system's trajectory and map.
• Object tracking and localization utilizing the YOLOv4 algorithm.
• Fusion of sensor measurements for comprehensive environmental understanding.

Main Results:

• The proposed system achieves real-time performance in localization and mapping.
• Accurate global positioning of the system and tracked objects is demonstrated.
• Successful mapping of unknown environments with integrated object tracking capabilities.

Conclusions:

• The fused sensor approach provides a robust solution for real-time SLAM and object localization.
• The system's ability to provide global positioning enhances its utility in critical applications.
• Potential applications include military operations, rescue, and disaster relief efforts.