Sensor-driven area coverage for an autonomous fixed-wing unmanned aerial vehicle

Area of Science:

• Robotics
• Aerospace Engineering
• Computer Vision

Background:

• Autonomous fixed-wing unmanned aerial vehicles (UAVs) are crucial for large-scale surveying.
• Non-holonomic dynamics and environmental disturbances pose challenges for UAV-based area coverage.
• Existing area coverage methods often rely on offline planning, assuming perfect trajectory following.

Purpose of the Study:

• To develop an adaptive area coverage strategy for fixed-wing UAVs that accounts for real-world flight dynamics.
• To enable in-situ path planning based on real-time sensor data and onboard camera models.
• To enhance the robustness and efficiency of autonomous aerial surveying missions.

Main Methods:

• An information-theoretic approach is employed to select headings that maximize expected information gain.
• A coverage map is maintained based on the UAV's actual trajectory and sensor data.
• The branch entropy concept is extended to ensure global coverage objectives are met.
• A projective camera model is used to compare expected and actual ground coverage.

Main Results:

• Hardware-in-the-loop simulations and real-world flight tests demonstrate the approach's effectiveness.
• The system successfully performs area coverage planning in situ, adapting to the UAV's actual pose.
• Integration with automatic takeoff and landing provides a fully automated aerial imagery survey platform.

Conclusions:

• The developed system offers a robust and efficient solution for autonomous area coverage with fixed-wing UAVs.
• This approach overcomes limitations of traditional offline planning by adapting to real-time conditions.
• The platform is well-suited for comprehensive aerial imagery surveys.