Non-Repetitive Scanning LiDAR Sensor for Robust 3D Point Cloud Registration in Localization and Mapping Applications

Area of Science:

• Robotics and Autonomous Systems
• Computer Vision
• Geospatial Data Processing

Background:

• 3D point cloud registration is crucial for autonomous navigation but faces challenges.
• Non-repetitive scanning LiDAR sensors offer new possibilities for 3D data acquisition.
• The feasibility of Spirograph-type LiDAR for robust registration requires investigation.

Purpose of the Study:

• To explore the feasibility of Spirograph-type non-repetitive scanning LiDAR for robust 3D point cloud registration.
• To propose and evaluate a novel registration method tailored to this unique sensor's scanning pattern.
• To assess the sensor's suitability for localization and mapping applications.

Main Methods:

• Characterization of data from a Spirograph-type non-repetitive scanning LiDAR sensor.
• Development of a registration method utilizing the sensor's unique scanning pattern to reconstruct and compare scans.
• Extraction of scene objects by comparing real and reconstructed scans to determine transformations.
• Performance evaluation on real datasets against state-of-the-art methods, with enhancements using loop closure and a Curve Fitting Derivative Filter (CFDT).

Main Results:

• The proposed method successfully registers successive 3D scans using the Spirograph-type scanning pattern.
• Performance improvements were achieved through loop closure constraints and CFDT for trajectory estimation.
• The sensor is demonstrated to be suitable for robust 3D point cloud registration.
• The method shows comparable accuracy to existing techniques but significantly faster processing times.

Conclusions:

• Spirograph-type non-repetitive scanning LiDAR is feasible for robust 3D point cloud registration.
• The proposed registration method offers an efficient solution for autonomous navigation.
• This sensor technology and registration approach advance the field of real-time localization and mapping.