Intelligent Real-Time Trajectory Optimization Framework for Multiple Time-Coordinated Hypersonic Glide Vehicles With No-Fly Zone Avoidance

An intelligent real-time trajectory optimization framework is proposed to address the time-coordinated guidance problem for multiple hypersonic glide vehicles with no-fly zone constraints in this paper. The proposed framework features a decoupled control architecture to manage the problem's complexity. Firstly, for the longitudinal channel, a homotopy-based optimization strategy is employed to generate angle of attack commands that satisfy terminal altitude and coordinated time-of-arrival constraints. Secondly, for the lateral channel, a hybrid situational awareness-driven segmented guidance strategy is developed, which utilizes a geometry-visibility-graph-based planner to generate a collision-free reference path, followed by a path-following controller and a situational awareness-driven terminal guidance law to ensure robust tracking and reliably steer the vehicles towards the designated target area. Thirdly, to achieve real-time performance, a modular set of dedicated deep neural networks (DNNs) is trained to act as surrogates for the computationally intensive optimization and prediction modules, enabling rapid, parallel inference and significantly accelerates the decision-making process. Finally, simulation results validate the effectiveness and robustness of the proposed framework, demonstrating that the DNN-enabled architecture significantly reduces computational latency in guidance command generation and thus paves the way for online, real-time operations.