Guoqing Shi

The interception of saturation incursions by Unmanned Aerial Vehicle (UAV) swarms presents critical challenges in multi-agent coordination, including the curse of dimensionality, heterogeneous interaction effects, and multi-scale decision-making requirements. This paper proposes a Hierarchical Multi-scale Mean-Field DDPG (HM-MF-DDPG) framework augmented by graph sampling and aggregation networks to address these challenges. The framework introduces three key innovations: (1) a graph-enhanced weighted mean-field approximation that employs attention mechanisms to dynamically assess the contextual importance of neighboring agents, overcoming the homogeneity limitation of conventional mean-field methods; (2) a hierarchical decision architecture that separates strategic coordination (via graph attention networks) from low-level flight control (via improved gated recurrent units with situational awareness modulation); and (3) a distributed target assignment mechanism formulated as a potential game and solved via parallel auction algorithms, enabling collision-free allocation without central coordination. Extensive simulations in a constructed UAV swarm interception environment demonstrate that the proposed framework achieves a 93% interception success rate with 50 interceptors against 25 intruders, outperforming Deep Deterministic Policy Gradient (DDPG) and Mean-Field DDPG (MF-DDPG) baselines in both convergence speed and task efficiency. The framework exhibits robust generalization across varying No-Fly Zone (NFZ) configurations and swarm scales, providing a scalable solution for cooperative interception under saturation incursions.

In the unmanned aerial vehicles (UAVs) flight control and navigation guidance system, trajectory prediction serves as a critical foundational component, with its accuracy and reliability directly influencing the system performance of the UAVs. However, existing research has predominantly focused on optimizing algorithm efficiency, failing to fully consider the impact of the UAV’s flight status on its trajectory. This has resulted in significant discrepancies between predicted results and actual trajectories in complex scenarios. Therefore, this paper proposes a trajectory prediction algorithm that integrates the UAVs’ behavioral intentions. Firstly, a behavioral intention recognition model is constructed using the Support Vector Machine (SVM) to accurately discriminate the UAV’s motion patterns and output the probability distribution of its future actions, thereby integrating semantic-level intention information into the prediction process. Secondly, the Bidirectional Gated Recurrent Unit (Bi-GRU) is employed to mine the spatial-temporal correlation features from trajectory data. Additionally, an attention mechanism is introduced to capture key information of sequence, enhancing the model’s ability to represent complex motion trends. The results of simulation experiments demonstrate that this algorithm exhibits significant advantages in terms of trajectory prediction accuracy and scene adaptability, providing more practical technical support for intelligent navigation and safety control of UAVs.

This study introduces a novel framework for intelligent unmanned BVR maneuver control within the context of adversarial games. The emphasis lies on three pivotal aspects: situational awareness, maneuver decision-making, and precise maneuver control. Within this paradigm, our unmanned aerial vehicles (UAVs) can assimilate crucial situational information through constructed situational vectors and execute sophisticated maneuvers, effectively addressing the intricacies of dynamic flight environments and various unpredictable scenarios within the game setting. To achieve granular maneuver control, this research introduces the Priority Heading Polling–Random Observation Weight (PHP-ROW) method, underpinned by deep reinforcement learning. This approach integrates two primary components: (1) the priority heading polling (PHP) mechanism, which governs the extent of flight trajectories while emphasizing heading control, and (2) the random observation weight (ROW) technique, which adeptly moderates the influence of roll angle rewards during the learning phase. The superiority of the PHP-ROW method is showcased by contrasting it against the conventional proximal policy optimization (PPO) algorithm. Conclusively, the utility and efficacy of the presented framework are corroborated through human–machine adversarial game simulations in a hyper-realistic environment. This investigation provides foundational theoretical and empirical contributions to the realm of intelligent unmanned aerial maneuver control, promising significant implications for the evolution of aviation technology in adversarial game contexts.

The Grey Wolf Optimizer (GWO) algorithm is recognized for its simplicity and ease of implementation, and has become a preferred method for solving global optimization problems due to its adaptability and search capabilities. Despite these advantages, existing Unmanned Aerial Vehicle (UAV) path planning algorithms are often hindered by slow convergence rates, susceptibility to local optima, and limited robustness. To surpass these limitations, we enhance the application of GWO in UAV path planning by improving its trajectory evaluation function, convergence factor, and position update method. We propose a collaborative UAV path planning model that includes constraint analysis and an evaluation function. Subsequently, an Enhanced Grey Wolf Optimizer model (NI–GWO) is introduced, which optimizes the convergence coefficient using a nonlinear function and integrates the Dynamic Window Approach (DWA) algorithm into the model based on the fitness of individual wolves, enabling it to perform dynamic obstacle avoidance tasks. In the final stage, a UAV path planning simulation platform is employed to evaluate and compare the effectiveness of the original and improved algorithms. Simulation results demonstrate that the proposed NI–GWO algorithm can effectively solve the path planning problem for UAVs in uncertain environments. Compared to Particle Swarm Optimization (PSO), Artificial Bee Colony (ABC), GWO, and MP–GWO algorithms, the NI–GWO algorithm can achieve the optimal fitness value and has significant advantages in terms of average path length, time, number of collisions, and obstacle avoidance capabilities.

Recently, developing multi-UAVs to cooperatively pursue a fast-moving target has become a research hotspot in the current world. Although deep reinforcement learning (DRL) has made a lot of achievements in the UAV pursuit game, there are still some problems such as high-dimensional parameter space, the ease of falling into local optimization, the long training time, and the low task success rate. To solve the above-mentioned issues, we propose an improved twin delayed deep deterministic policy gradient algorithm combining the genetic algorithm and maximum mean discrepancy method (GM-TD3) for multi-UAV cooperative pursuit of high-speed targets. Firstly, this paper combines GA-based evolutionary strategies with TD3 to generate action networks. Then, in order to avoid local optimization in the algorithm training process, the maximum mean difference (MMD) method is used to increase the diversity of the policy population in the updating process of the population parameters. Finally, by setting the sensitivity weights of the genetic memory buffer of UAV individuals, the mutation operator is improved to enhance the stability of the algorithm. In addition, this paper designs a hybrid reward function to accelerate the convergence speed of training. Through simulation experiments, we have verified that the training efficiency of the improved algorithm has been greatly improved, which can achieve faster convergence; the successful rate of the task has reached 95%, and further validated UAVs can better cooperate to complete the pursuit game task.

This paper proposes a method based on a Bayesian network model to study the intelligent tactical decision making of formation coordination. For the problem of formation coordinated attack target allocation, a coordinated attack target allocation model based on the dominance matrix is constructed, and a threat degree assessment model is constructed by calculating the minimum interception time. For the problem of real-time updating of the battlefield situation in the formation confrontation simulation, real-time communication between the UAV formation on the battlefield is realized, improving the efficiency of communication and target allocation between formations on the battlefield. For the problem of UAV autonomous air combat decision making, on the basis of the analysis of the advantage function calculation of the air combat decision-making model and a Bayesian network model analysis, the network model's nodes and states are determined, and the air combat decision-making model is constructed based on the Bayesian network. Our formation adopts the Bayesian algorithm strategy to fight against the blue side's UAVs, and the formation defeats the blue UAVs through coordinated attack, which proves the reasonableness of coordinated target allocation. An evaluation function is established, and the comprehensive scores of our formation are compared with those of other algorithms, which proves the accuracy and intelligibility of the decision making of the Bayesian network.

In this paper, a LabVIEW-based online monitoring and safety evaluation system for UAVs is designed to address the deficiencies in UAV flight state parameter monitoring and safety evaluation. The system consists of a lower unit for UAV recording and an upper unit on the ground. The lower unit collects and detects flight data and connects to the upper unit through a wireless digital transmission module via a serial port. The upper unit receives the data and carries out the monitoring and safety situation evaluation of the UAV. The lower unit of the system adopts multi-sensors to collect UAV navigation information in real time to achieve flight detection, while the upper unit adopts LabVIEW to design the UAV online monitoring and safety situation prediction system, enabling monitoring and safety situation prediction during UAV navigation. The test results show that the system can detect and comprehensively display the navigation information of the UAV in real time, and realize the safety evaluation and warning function of the UAV.

With the burgeoning impact of artificial intelligence on the traditional UAV industry, the pursuit of autonomous UAV flight has emerged as a focal point of contemporary research. Addressing the imperative for advancing critical technologies in autonomous flight, this paper delves into the realm of UAV flight state recognition and trajectory prediction. Presenting an innovative approach focused on improving the precision of unmanned aerial vehicle (UAV) path forecasting via the identification of flight states, this study demonstrates its efficacy through the implementation of two prediction models. Firstly, UAV flight data acquisition was realized in this paper by the use of multi-sensors. Finally, two models for UAV trajectory prediction were designed based on machine learning methods and classical mathematical prediction methods, respectively, and the results before and after flight pattern recognition are compared. The experimental results show that the prediction error of the UAV trajectory prediction method based on multiple flight modes is smaller than the traditional trajectory prediction method in different flight stages.

In response to the challenges of estimating missile launch timing during close-range unmanned autonomous air combat in the future, this paper proposes an autonomous attack decision-making method based on hierarchical virtual Bayesian reinforcement learning (HVBRL). First, a six-degree-of-freedom (6-DOF) high-fidelity aircraft dynamics model along with missile dynamics and guidance rate models are constructed. Second, the HVBRL algorithm is introduced, where the low-level algorithm outputs control parameters and the high-level algorithm generates control commands. Given that the number of missile hits on a target under specific conditions follows a binomial distribution, a simple prior knowledge can be introduced through its conjugate prior, the Beta distribution, to avoid prolonged exploration of ineffective areas. Moreover, carrying only a limited number of missiles and predicting the number of hits by multiple virtual missiles in specific states through a neural network circumvents the computational complexity issue associated with carrying an excessive number of missiles. Finally, this paper presents the low-level training algorithm, the high-level training algorithm, and the high-level self-play training algorithm. Experimental results show that our method significantly reduces the simulation computational complexity. Compared with the Monte Carlo method carrying 1000 missiles, the simulation speed of the high-level training algorithm is increased by 32.75 times, and that of the high-level self-play algorithm is increased by 23 times. Moreover, the estimated missile hit probability with bias can effectively guide the timing of missile launches in close-range air combat, which has significant implications for intelligent autonomous air combat decision-making and operational analysis.

With the impact of artificial intelligence on the traditional UAV industry, autonomous UAV flight has become a popular research field at present. Based on the demand for research on key technologies for autonomous UAV flight, this paper studies UAV flight state recognition. This paper is based on multi-sensor acquisition of UAV on-board information, and uses the collected information for data fusion to complete UAV flight state identification. Firstly, UAV flight data acquisition and data preprocessing are carried out; secondly, UAV flight trajectory features are extracted based on multidimensional data fusion; finally, UAV flight state recognition model based on PCA-DAGSVM model is established. The results show that the algorithm based on multi-sensor data fusion has good recognition effect in the UAV flight state recognition problem. The recognition accuracy of the algorithm exceeds that of the random forest model, and its accuracy in the training set of UAV flight state recognition is more than 90%, and its accuracy in the test set is more than 80%.