UAV Visual and Laser Sensors Fusion for Detection and Positioning in Industrial Applications

Area of Science:

• Robotics and Automation
• Computer Vision
• Sensor Fusion

Background:

• Accurate localization of Unmanned Autonomous Vehicles (UAVs) is crucial for inspection and maintenance tasks in industrial and civilian infrastructures.
• Existing methods often struggle with precise positioning relative to cylindrical structures like pipes.
• The integration of diverse sensor data, such as vision and laser, presents an opportunity to improve localization accuracy.

Purpose of the Study:

• To develop and present a robust solution for localizing UAVs with respect to pipes and other cylindrical elements.
• To combine the strengths of vision and laser-based sensors for enhanced positioning accuracy.
• To enable precise robotic navigation for inspection and maintenance operations.

Main Methods:

• Utilized a probabilistic RANSAC-based algorithm to segment cylindrical features from laser scan data.
• Employed computer vision techniques, seeded by laser scan segmentation, to accurately determine robot position.
• Integrated laser scan registration priors to define cylinder apparent contours for visual pose estimation.

Main Results:

• Successfully demonstrated accurate localization of UAVs relative to cylindrical structures.
• The sensor fusion approach effectively leveraged complementary features from vision and laser sensors.
• The system achieved precise visual estimation of cylinder pose through degenerate quadratic conic estimation.

Conclusions:

• The proposed system offers a reliable solution for UAV localization in pipe-centric inspection and maintenance.
• Combining laser-based segmentation with vision-based pose estimation significantly improves positioning accuracy.
• This approach enhances the capabilities of autonomous robots in complex infrastructure environments.