Autonomous UAV Landing and Collision Avoidance System for Unknown Terrain Utilizing Depth Camera with Actively...

Area of Science:

• Robotics and Control Systems
• Computer Vision for Autonomous Systems
• Aerospace Engineering

Background:

• Autonomous landing is critical for unmanned aerial vehicles (UAVs).
• Existing methods (lidar, cameras, beacons) have limitations like restricted perception or pre-defined landing zones.
• RGB-D sensors offer combined visual and depth data but often have a limited field of view.

Purpose of the Study:

• To develop a novel autonomous landing system for UAVs.
• To overcome the limitations of current landing strategies, particularly field of view and obstacle perception.
• To enable safe, obstacle-aware low-altitude flight and landing.

Main Methods:

• Utilized an RGB-D camera mounted on an actively controlled gimbal for enhanced field of view.
• Implemented a combined UAV and gimbal motion strategy for obstacle detection and avoidance.
• Integrated the system with PX4 flight stack, CubeOrange flight controller, and Jetson nano onboard computer.

Main Results:

• Demonstrated successful static testing on a real vehicle.
• Validated the system architecture's correctness through simulation and static tests.
• Showcased the potential for real-world deployment of the autonomous landing system.

Conclusions:

• The proposed gimbal-actuated RGB-D camera system effectively expands the field of view for autonomous landing.
• The combined motion strategy enables robust static obstacle detection and avoidance during low-altitude flight.
• The system architecture is validated, indicating feasibility for practical UAV applications.