GPGPU accelerated real-time potential field based formation control for Unmanned Aerial Vehicles

One of the research topics for the autonomous UAVs is formation flight control that provides two or more aircraft flies together under a discipline. For aerial vehicles the advantages of performing formation flight include fuel saving, improved efficiency in air traffic control and cooperative task allocation. Besides the benefits of the autonomous formation flight it has some difficulties like providing collision avoidance in narrow flight zone. Furthermore an autonomous formation flight must provide dynamic routines like getting formation position or changing formation schemes. Potential field based autonomous control is one of the commonly used control techniques providing dynamic and precise control. The bottleneck for the potential field based control is computation power needs especially for global path planning for dynamic, high resolution and large sized fields. GPGPU is one of the parallel computing architectures that provide programming massively parallel SIMD applications on GPUs. In this work GPGPU accelerated real-time potential field based formation control approach is designed and examined under simulation environment for UAVs. In order to reveal the precise and dynamic control features high-resolution potential field based global path planning models have been developed in real-time successfully.