31 May 2019 in IEEE Transactions on Aerospace and Electronic Systems
This study presents a method to adjust the waypoints of an unpowered air vehicle to compensate for the influence of wind on the trajectory. A framework combining pre-flight waypoint planning and in-flight waypoint adjustment is proposed. In the offline planning phase, optimal trajectories under various wind profile combinations are generated by using a direct optimization method. Waypoints are extracted from the obtained trajectories for each wind condition. Then, deviations of each waypoint due to wind from the corresponding waypoint on the trajectory for a zero-wind case are obtained; these deviations are used to construct the models of waypoint deviation as functions of wind speed and direction via least-squares regression. In the online adjustment phase, the wind-compensated waypoint is computed using the waypoint deviation model and the estimated wind velocity. A nonlinear six degrees-of-freedom simulation, incorporating a guidance and control system and a realistic wind profile, is performed to demonstrate the effectiveness of the proposed waypoint management framework.
22 July 2020 in Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering
An impact angle control guidance law is proposed for stationary target interception considering missile's field-of-view limit and speed changes. The proposed impact angle control guidance law is structured as a biased proportional navigation with a time-varying bias. The proposed guidance law does not involve any switching logic for maintaining lock-on; hence, the guidance command is continuous during the entire engagement. Unlike the most existing studies, the proposed method guarantees that the impact angle error converges to zero before interception without the constant-speed assumption. To realize these desirable properties, the positive invariance of the bounded look angle interval and the change of independent variable are utilized. Numerical simulations are conducted to demonstrate the performance of the proposed guidance law.
20 July 2022 in IEEE Transactions on Neural Networks and Learning Systems
Parameterized max-affine (PMA) and parameterized log-sum-exp (PLSE) networks are proposed for general decision-making problems. The proposed approximators generalize existing convex approximators, namely max-affine (MA) and log-sum-exp (LSE) networks, by considering function arguments of condition and decision variables and replacing the network parameters of MA and LSE networks with continuous functions with respect to the condition variable. The universal approximation theorem (UAT) of PMA and PLSE is proved, which implies that PMA and PLSE are shape-preserving universal approximators for parameterized convex continuous functions. Practical guidelines for incorporating deep neural networks within PMA and PLSE networks are provided. A numerical simulation is performed to demonstrate the performance of the proposed approximators. The simulation results support that PLSE outperforms other existing approximators in terms of a minimizer and optimal value errors with scalable and efficient computation for high-dimensional cases.
1 March 2024 in Journal of Aerospace Engineering
26 May 2021 in IEEE Access
A data-driven path replanning algorithm based on Pythagorean-hodograph (PH) curve is proposed for unpowered gliding vehicles considering the terminal velocity constraint in uncertain windy environments. Based on the characteristics of the PH curve, the terminal velocity constraint is satisfied by adjusting path parameters including the arc length in path replanning. For the path replanning, a tailored PH curve regeneration algorithm is proposed to resolve the inconsistent path replanning issue of the existing PH curve generation method. An artificial-neural-network-based path replanner is trained to provide the path parameters corresponding to the terminal velocity constraint. Unlike most model-based methods, the proposed method can deal with the external disturbance by replanning the flight path. Compared with other data-driven studies, the proposed method does not require a computationally expensive trajectory optimization process to collect training data. Performance evaluation and comprehensive comparison are performed via numerical simulation considering various types of modeled and unmodeled wind uncertainties using several network configurations.