Traversable Region Detection and Tracking for a Sparse 3D Laser Scanner for Off-Road Environments Using Range Images

Area of Science:

• Robotics
• Computer Vision
• Autonomous Systems

Background:

• Unmanned ground vehicle (UGV) navigation in off-road environments is complex due to unpredictable terrain.
• Accurate detection and tracking of traversable regions are critical for safe and efficient UGV operation.
• Existing methods may struggle with the dynamic and unstructured nature of off-road conditions.

Purpose of the Study:

• To develop and validate a robust method for detecting and tracking traversable regions for UGVs operating in off-road conditions.
• To enhance UGV navigation safety and efficiency in challenging outdoor environments.
• To leverage 3D laser scanner data for real-time traversable area identification.

Main Methods:

• A range-image-based approach for processing 3D laser scanner point cloud data to detect traversable regions.
• Implementation of a Bayesian fusion algorithm to track identified traversable regions across consecutive sensor frames.
• Utilizing real-world data from UGV operations on rough terrain for method validation.

Main Results:

• The proposed range-image-based method efficiently processes 3D point cloud data.
• The Bayesian fusion algorithm effectively tracks traversable regions over time.
• Demonstrated effectiveness in identifying safe driving paths for UGVs in complex off-road scenarios.

Conclusions:

• The developed method significantly improves the ability of UGVs to navigate challenging off-road terrains.
• The approach offers a reliable solution for enhancing UGV autonomy and operational safety.
• This technique shows promise for advancing robotic navigation in unstructured environments.