CTFF: An Efficient Strategy for Continuously Targeting Fire Fighting with Quadrotor

TL;DR: In this article , the Gauss pseudo-pectral method is utilized to transform the optimal control problem into a nonlinear constrained quadratic programming problem, and then the control input can be generated analytically.

Abstract: This paper considers the optimal control problem of continuously targeting for extinguishing a moving fire source with a quadrotor. It aims to enhance computational time-efficiency, and real-time targeting performance compared with generic optimizers onboard, even suffering from severe nonlinearities. To this end, an efficient strategy is proposed for this optimal control problem. First, the Gauss Pseudospectral Method is utilized to transform the optimal control problem into a nonlinear constrained quadratic programming problem. Then, via the Sequential Quadratic Programming (SQP) algorithm, the control input can be generated analytically. To further improve the computational efficiency, the techniques, separation of variables and prior gradient calculation, are applied. Through comparative simulations, it is verified that the proposed method has excellent targeting performance and high computing efficiency. Further analysis demonstrate that the proposed efficient strategy can complete the control plan in approximately 0.5 ms, which is 20 times faster than the generic optimizers.