Based on the versatility and effectiveness of the siamese neural network, the technology of unmanned aerial vehicle visual object tracking has found widespread application in various fields including military reconnaissance, intelligent transportation, and visual positioning. However, due to complex factors, such as occlusions, viewpoint changes, and interference from similar objects during UAV tracking, most existing siamese neural network trackers struggle to combine superior performance with efficiency. To tackle this challenge, this paper proposes a novel SiamSTM tracker that is based on Slight Aware Enhancement Transformer and Multiple matching networks for real-time UAV tracking. The SiamSTM leverages lightweight transformers to encode robust target appearance features while using the Multiple matching networks to fully perceive response map information and enhance the tracker’s ability to distinguish between the target and background. The results are impressive: evaluation results based on three UAV tracking benchmarks showed superior speed and precision. Moreover, SiamSTM achieves over 35 FPS on NVIDIA Jetson AGX Xavier, which satisfies the real-time requirements in engineering.

UAV visual-object-tracking technology based on Siamese neural networks has great scientific research and practical application value, and it is widely used in geological surveying, reconnaissance monitoring, and environmental monitoring. Due to the limited onboard computational resources and complex real-world environments of drones, most of the existing tracking systems based on Siamese neural networks struggle to combine excellent performance with high efficiency. Therefore, the key issue is to study how to improve the accuracy of target tracking under the challenges of real-time performance and the above factors. In response to this problem, this paper proposes a real-time UAV tracking system based on interframe saliency transformer and lightweight multidimensional attention network (SiamITL). Specifically, interframe saliency transformer is used to continuously perceive spatial and temporal information, making the network more closely related to the essence of the tracking task. Additionally, a lightweight multidimensional attention network is used to better capture changes in both target appearance and background information, improving the ability of the tracker to distinguish between the target and background. SiamITL is effective and efficient: extensive comparative experiments and ablation experiments have been conducted on multiple aerial tracking benchmarks, demonstrating that our algorithm can achieve more robust feature representation and more accurate target state estimation. Among them, SiamITL achieved success and accuracy rates of 0.625 and 0.818 in the UAV123 benchmark, respectively, demonstrating a certain level of leadership in this field. Furthermore, SiamITL demonstrates the potential for real-time operation on the embedded platform Xavier, highlighting its potential for practical application in real-world scenarios.

This paper presents an interval type-2 fuzzy proportional–integral–derivative (IT2F-PID) controller that is designed using a new disassembled gradational optimization (D-GO) method. A PID controller is first optimized using the D-GO method and then connected to a type-1 fuzzy logic system (T1-FLS). The parameters of the T1-FLS are optimized, and the T1-FLS is blurred into the interval type-2 fuzzy logic system (IT2-FLS). Finally, the IT2F-PID controller is formed. The proposed method is compared with the concurrent and general optimization methods. The simulation results show that the D-GO method reduces the optimization time by over 90% compared with the general method, and decreases the integral-of-time-absolute-error (ITAE) by 30%. Beyond that, compared with the concurrent optimization method, the D-GO method reduces time by over 25%, and the ITAE value by about 95%. In the normal case, model uncertainty, target uncertainty, and external disturbance, the control ability of the IT2F-PID controller designed using the D-GO method is verified via simulations using a nonlinear forced closed-loop system. The results show that the overshoot is reduced by 80% and the fluctuation is reduced by 67% compared with a traditional PID controller and an IT2F-PID controller built using the general method.

Point cloud registration is a critical problem because it is the basis of many 3D vision tasks. With the popularity of deep learning, many scholars have focused on leveraging deep neural networks to address the point cloud registration problem. However, many of these methods are still sensitive to partial overlap and differences in density distribution. For this reason, herein, we propose a method based on rotation-invariant features and using a sparse-to-dense matching strategy for robust point cloud registration. Firstly, we encode raw points as superpoints with a network combining KPConv and FPN, and their associated features are extracted. Then point pair features of these superpoints are computed and embedded into the transformer to learn the hybrid features, which makes the approach invariant to rigid transformation. Subsequently, a sparse-to-dense matching strategy is designed to address the registration problem. The correspondences of superpoints are obtained via sparse matching and then propagated to local dense points and, further, to global dense points, the byproduct of which is a series of transformation parameters. Finally, the enhanced features based on spatial consistency are repeatedly fed into the sparse-to-dense matching module to rebuild reliable correspondence, and the optimal transformation parameter is re-estimated for final alignment. Our experiments show that, with the proposed method, the inlier ratio and registration recall are effectively improved, and the performance is better than that of other point cloud registration methods on 3DMatch and ModelNet40.

Visual object tracking is a key technology that is used in unmanned aerial vehicles (UAVs) to achieve autonomous navigation. In recent years, with the rapid development of deep learning, tracking algorithms based on Siamese neural networks have received widespread attention. However, because of complex and diverse tracking scenarios, as well as limited computational resources, most existing tracking algorithms struggle to ensure real-time stable operation while improving tracking performance. Therefore, studying efficient and fast-tracking frameworks, and enhancing the ability of algorithms to respond to complex scenarios has become crucial. Therefore, this paper proposes a cross-parallel attention and efficient match transformer for aerial tracking (SiamEMT). Firstly, we carefully designed the cross-parallel attention mechanism to encode global feature information and to achieve cross-dimensional interaction and feature correlation aggregation via parallel branches, highlighting feature saliency and reducing global redundancy information, as well as improving the tracking algorithm’s ability to distinguish between targets and backgrounds. Meanwhile, we implemented an efficient match transformer to achieve feature matching. This network utilizes parallel, lightweight, multi-head attention mechanisms to pass template information to the search region features, better matching the global similarity between the template and search regions, and improving the algorithm’s ability to perceive target location and feature information. Experiments on multiple drone public benchmark tests verified the accuracy and robustness of the proposed tracker in drone tracking scenarios. In addition, on the embedded artificial intelligence (AI) platform AGX Xavier, our algorithm achieved real-time tracking speed, indicating that our algorithm can be effectively applied to UAV tracking scenarios.