Concerned with the problem of interceptor midcourse guidance trajectory online planning satisfying multiple constraints, an online midcourse guidance trajectory planning method based on deep reinforcement learning (DRL) is proposed. The Markov decision process (MDP) corresponding to the background of a trajectory planning problem is designed, and the key reward function is composed of the final reward and the negative step feedback reward, which lays the foundation for the interceptor training trajectory planning method in the interactive data of a simulation environment; at the same time, concerned with the problems of unstable learning and training efficiency, a trajectory planning training strategy combined with course learning (CL) and deep deterministic policy gradient (DDPG) is proposed to realize the progressive progression of trajectory planning learning and training from satisfying simple objectives to complex objectives, and improve the convergence of the algorithm. The simulation results show that our method can not only generate the optimal trajectory with good results, but its trajectory generation speed is also more than 10 times faster than the hp pseudo spectral convex method (PSC), and can also resist the error influence mainly caused by random wind interference, which has certain application value and good research prospects.

For traditional predictor-corrector guidance algorithm for reentry glide vehicle, it cost a lot of time to obtain predicted flight range with a slow speed to iterate. In this paper, according to residual network (ResNet)’s block and dynamic model of vehicle, through analyzing the characteristics of predicted flight range with constraints, the flight range prediction block and flight range prediction neural network are designed, which can obtain the predicted range accurately and quickly; then aiming at the separation between guidance logic and no-fly zone avoidance logic, which may lead to guidance failure and increasing of the sign variation number of the bank angle, the no-fly zone crossrange and the no-fly zone mapping crossrange are proposed in this paper. According the repulsion force of artificial potential field, an adaptive crossrange corridor combining guidance logic and no-fly zone avoidance logic is proposed, and the convergence of the corridor is analyzed theoretically. Through simulation, the block number of flight range prediction network is determined firstly. By this method, the efficiency of lateral guidance can be improved. Then, through the simulation with the different no-fly zones under different disturbed conditions, the stability and validity of the guidance method are verified. Finally, compared with other predictor-corrector algorithms, the proposed method can realize guidance with less sign variation number of bank angle and better avoidance for no-fly zones.

Aimed at the problem of the insufficient utilization of intention information and flight data for the trajectory prediction of a reentry glide target, an intelligent trajectory prediction algorithm based on intention inference was proposed. Firstly, a control parameter prediction network was developed to predict the variation of the control parameter values. Secondly, based on the Gaussian mixture model and Dubins circle, the importance of strategic places was modeled, the functional relationship between the strategic places and trajectories was established, and the influence factors of target threat were analyzed. Subsequently, an intention inference network was designed, which can accurately identify the key point the target intends to attack and realize the nonlinear calculation of the target’s threat value. Finally, according to the common guidance law of reentry glide target, the lateral sign-variation rule was designed to carry out the target trajectory prediction based on intention inference. Simulation revealed that the parameter prediction network designed in this paper can realize the modification of filtering control parameters and effectively predict the value of these parameters. Moreover, taking the result of intention inference network, trajectory prediction in different prediction cases was accomplished. The maximal error of spatial distance (MESD) of the proposed method was less than 9 km when the prediction time was 100 s, and the best result was obtained in the long-term prediction. Compared with other mainstream prediction methods, the proposed method obtained the best prediction accuracy.

Optimal trajectory generation for the guidance of re-entry glide vehicles is of great significance. To realize a faster generation speed and consistency with the guidance mechanism, an improved convex optimization trajectory generation algorithm based on the flight range domain for the re-entry glide vehicles is proposed in this paper. Firstly, according to the definition of the range-to-go, the projected range-to-go of the re-entry glide vehicle is presented when the dynamic model is converted to the flight range domain. Then, the attack angle and bank angle are expanded to the state variables and the change rate, which is designed as a new control variable. When the dynamic models and constraints are convexificated and discretized, the vehicle trajectory discrete convex model in the flight range domain is proposed. In order to further improve the generation speed and accuracy, an initial trajectory generation method that is close to the guidance requirements is proposed by the landing points of different control laws. In addition, by analyzing the nonlinear illegal degree of grid points, the probability density of grid points and the adjustment strategy of grid points are proposed. Finally, the ablation experiment shows that the initial trajectory generation and distributed grid points method works. With different target points, different no-fly zones, different initial states, and the Monte Carlo experiment, our method can effectively and robustly complete the generation. The lateral and longitudinal generation error is less than 1 km. And compared with the Gaussian pseudo-spectral method, our method obtained comparable accuracy and faster speed.

To address the challenges of detection and acquisition caused by trajectory prediction errors during the midcourse–terminal guidance handover phase in cross-domain unmanned aerial vehicles (CDUAV), this study proposes a collaborative multi-interceptor detection coverage optimization method based on predictive error pipeline modeling. Firstly, we employ nonlinear least squares to fit parameters for the motion model of CDUAV. By integrating error propagation theory, we derive a recursive expression for error pipelines under t-distribution and establish a parametric model for the target’s high-probability region (HPR). Next, we analyze target acquisition scenarios during guidance handover and reformulate the collaborative detection problem as a field-of-view (FOV) coverage optimization task on a two-dimensional detection plane. This framework incorporates the target HPR and the seeker detection FOV models, with an objective function defined for coverage optimization. Finally, inspired by wireless sensor network (WSN) coverage strategies, we implement the starfish optimization algorithm (SFOA) to enhance computational efficiency. Simulation results demonstrate that compared to Monte Carlo statistical methods, our parametric modeling approach reduces prediction error computation time from 15.82 s to 0.09 s while generating error pipeline envelopes with 99% confidence intervals, showing superior generalization capability. The proposed collaborative detection framework effectively resolves geometric coverage optimization challenges arising from mismatches between target HPR and FOV morphology, exhibiting rapid convergence and high computational efficiency.