Target recognition and tracking based on multi-rotor UAVs have the advantages of low cost and high flexibility. It can monitor low-altitude targets with high intensity. It has great application prospects in national defense, military, and civil fields. The existing algorithms for aerial small target recognition and tracking have the disadvantages of slow speed, low accuracy, poor robustness, and insufficient intelligence. Aiming at the problems of existing algorithms, this paper first makes a lightweight improvement for the YOLOv4 network recognition algorithm suitable for small target recognition and tests it on the VisDrone dataset. The accuracy of the improved algorithm is increased by 1.5% and the speed is increased by 3.3 times. Then, by analyzing the response value, the KCF tracking situation is judged, and the template update of the adaptive learning rate is realized. When the tracking fails, the target is re-searched and tracked based on the recognition results and the similarity judgment. Finally, experiments are carried out on the multi-rotor UAV, and the adaptive zoom tracking strategy is designed to track pedestrians, cars, and UAVs. The results show that the proposed algorithm can achieve stable tracking of long-distance small targets.

This paper considers a reconnaissance task allocation problem for multiple unmanned aerial vehicles (UAVs) in 3D urban environments. In this paper, we present an extended heterogeneous targets reconnaissance task allocation model which introduced cuboid targets for 3D urban environment to improve the fidelity of the model. A reconnaissance method is designed for each type of target, and the mission is described as a heterogeneous target multi-traveling salesman problem model for solving complex optimization problems with multiple constraints.To address these complex optimization problems, multi-group symbiotic organisms search algorithms (MGSOS) are proposed, which maintain the diversity of species in the population through multi-group strategies and enhance information exchange between individuals in three stages. Real-number encoding is used to satisfy partial constraints and simplify the search space, improving the optimization efficiency of the solution. The simulation results show that the MGSOS algorithm can consider the characteristics of UAV sensor performance and heterogeneous targets. It outperforms the common symbiotic organisms search (SOS) algorithm in terms of the optimality of assignment results, and is suitable for larger scale urban reconnaissance task allocation problems.

Infrared and visible image fusion technology has a wide range of applications in many fields such as target detection and tracking. Existing image fusion methods often overlook the scale hierarchical structure information of features, with local and global features not being closely interconnected. Typically, improvements focus on the network structure and loss functions, while the intimate connection between the quality of the source images and the feature extraction network is often neglected. The aforementioned issues lead to artifacts and blurring of fused images; besides, the detailed edge information can not be well reflected. Therefore, a method of infrared and visible image fusion based on a generative adversarial network (GAN) with multi-level detail enhancement is proposed in this paper. Firstly, the edge information of the input source image is enriched by the multi-level detail enhancement method, which improves the image quality and makes it more conducive to the learning of feature extraction network. Secondly, the residual-dense and multi-scale modules are designed in the generator and the connection between local and global features is established to ensure the transmissibility and coherence of the feature information. Finally, by designing the loss function and dual discriminator constraints to constrain the fusion image, more structure and detail information are added in continuous confrontation. The experimental results show that the fused image contains more detailed texture information and prominent thermal radiation targets. It also outperforms other fusion methods in terms of average gradient (AG), spatial frequency (SF) and edge intensity (EI) metrics, with values surpassing the sub-optimal metrics of 65.41%, 65.09% and 55.22%, respectively.

In recent years, Wireless Sensor Networks (WSNs) have seen increasing application in GPS-Denied environments, such as indoor localization and mobile robot navigation. Among various positioning algorithms, the Least Squares Method is widely used for its simplicity and robustness. While numerous optimization algorithms have been developed to reduce localization error, comprehensive studies on the sources and propagation of positioning errors remain limited.This paper focuses on the error propagation mechanisms in range-based WSN localization. It conducts an in-depth analysis of the primary sources of positioning error, including ranging errors, anchor node position errors, and the influence of network topology on error propagation. A novel, simplified error propagation model based on covariance theory is proposed to quantitatively estimate the cumulative impact of these errors on localization accuracy.To evaluate the accuracy and practicality of the proposed algorithm, comprehensive assessments were carried out through both MATLAB simulations and physical experiments in representative indoor environments. The results show that the average discrepancy between the estimated and actual positioning errors remains within a few centimeters, demonstrating strong predictive capability. Furthermore, the Geometric Dilution of Precision is introduced to assess the influence of base station configurations. The experimental results further confirm that the proposed algorithm maintains strong robustness even in suboptimal geometric configurations of the network. In conclusion, the error propagation model proposed in this study not only significantly simplifies the theoretical computation of localization errors but also offers high accuracy in error estimation, making it particularly suitable for multi-sensor fusion scenarios that require precise error modeling. It also provides new insights and quantitative tools for optimizing WSN structures and improving overall localization performance.

A method for determination of attitude of the airborne remote sensing platform is proposed. In this method, the attitude of airborne platform can be obtained on‐line by following four steps: (i) recording the attitude of the platform at time t ₀ , which serves as the initial attitude; (ii) starting from t ₀ , taking aerial images using the dedicated camera fixed on the airborne platform at time t _k ( k = 0, 1, 2, …) regularly and noting as the k th ( k = 0, 1, 2, …) aerial image; (iii) calculating the attitude change between every two adjacent images; (iv) determining the attitude of the platform at a time according to the initial attitude and the related attitude change. Simulation results proves the theoretical feasibility of the method. The attitude error sources are also be analysed. The attitude accuracy is proportional to focal length, field angle and matching accuracy of the corresponding points. Physical experiments were carried out to illustrate the feasibility and effectiveness of the proposed method.

This paper presents a fully integrated, end-to-end system framework for multimodal robots navigating complex ground, air, and ceiling environments. By seamlessly synthesizing perception, planning, and control architectures, the system employs a 3D LiDAR SLAM scheme using FAST-LIO2 and Patchwork++ for adaptive segmentation. To address stringent spatial safety constraints without computationally exhaustive dynamic modeling, we introduce a unified cost function incorporating heuristic spatial penalties for ground steering, flight energy, and ceiling-proximity operations. The generated trajectories are refined via B-spline optimization for smoothness. Furthermore, a robust seven-state finite state machine (FSM) coupled with a hierarchical tracking controller optimally manages autonomous mode transitions, ensuring continuous operational flow across diverse locomotion domains. Extensive simulations and real-world physical experiments—including underground pipeline inspections—demonstrate the framework's superiority. In evaluated complex constrained scenarios, the proposed system outperforms the widely recognized TABV baseline, reducing trajectory length by up to 9.09% and execution time by 10.87%. Additionally, dedicated low-speed tracking evaluations confirm the controller's precision, maintaining tracking errors consistently within a 0.1m margin. These results validate the system's high efficiency, reliability, and practical deployability in constrained 3D environments.