High-performance graphics processing unit-based strategy for tuning a unmanned aerial vehicle controller subject to time-delay constraints

Recently, high-performance computing strategies have been implemented to improve performance analysis and reduce the development time of new solutions in robotic applications, such as path planning, machine learning, and vision, which require massive matrix computations. In this sense, this work aims to study the aerial robots' behavior during their mission execution. Due to the large search space in the set of parameter combinations and the high computational cost required to perform such an analysis after sequentially executing thousands of simulations, this work proposes an open-source graphics processing unit (GPU)-based implementation to simulate the robot behavior. A GPU-accelerated flight route analysis for multi-unmanned aerial vehicle (UAV) systems is proposed for the tuning control problem in the parameters' space considering the problem of delay in sending information to a ground control station. Considering our implementation, the experimental results show a speedup up to 325x$$ \times $$, 629x$$ \times $$, and 5959x$$ \times $$ in comparison to the parallel version with 16 threads, C coder converter, and native Matlab code, respectively. The implementation is available in the Colab Google platform and it can easily be expanded for analyses involving larger amounts of different parameters, robot models, strategies, and controllers.