Autonomous Air Refueling Path Planning for UAVs with ArrayFire

John Melonakos Case Studies Leave a Comment

Researchers from the Aeronautics and Space Technologies Institute of the Turkish Air Force Academy credit ArrayFire in a paper published in the Journal of Intelligent & Robotic Systems. The paper is titled “Sigmoid Limiting Functions and Potential Field Based Autonomous Air Refueling Path Planning for UAVs” and showcases ArrayFire in a real-time application of UAV path planning.


Summary

This paper builds on previous approaches for autonomous air-refueling (AAR) path planning for Unmanned Aerial Vehicles (UAVs). Deficiencies from previous approaches, like smooth maneuvers in the tanker approach and the boundary functions of the potential zones, have been handled. Furthermore, special pattern parameters are added to the approach which makes it suitable for different kind of UAVs that has variable flight speed and turn radius parameters.

Important originality of the approach comes from using sigmoid limiting functions while modeling dynamic behaviors of the potential fields that are based on path planning algorithms. The replacement of logical binary boundary functions with sigmoid limiting functions creates a heavy computational burden. In order to use the AAR path planning approach in a real-time application, the computation is performed in Graphical Processing Units (GPUs) based on parallel architectures with ArrayFire.

The Results

In Figure 1 below, a diagram for the application is shown. The tanker flies in an orbit that the UAV autonomously must find in order to refuel.

In Figure 7, we see the difference between the previous versions with binary membership functions in (a) and the improved version with the smoother sigmoid boundary function in (b).

In Figure 12, we see the result of the improved version is a shorter, smoother path for the UAV trajectory to find the refueling tanker.

Performance Results

ArrayFire’s impact on this algorithm can be seen from the performance benchmarks provided in Figure 11. The Parallel Computing Performace comes from ArrayFire using an NVIDIA GPU.


Thanks to these researchers for sharing their great work with us!

Leave a Reply

Your email address will not be published. Required fields are marked *