Multispectral object detection has attracted increasing attention recently due to its superior detection capacity under various illumination conditions. The key challenge lies in the effective aggregation of multi-spectral features to derive highly discriminative representations. To address this challenge, we propose a novel Multidimensional Fusion Network (MMFN) to explore multi-modal information from local, global, and channel perspectives. Specifically, at the local level, local features of different modalities and their inter-relationships are captured by a window-shifted fusion. As a complement to the local information, we designed a global interaction module that facilitates the fusion of holistic, high-level semantic information spanning the entire image. We distillate the channel dependencies and complementarities between different modalities through cross-channel learning and generate the final fused representation. Comprehensive experiments conducted on three publicly available datasets provide compelling evidence validating the superiority of the proposed methodology. The results exhibit notable performance gains over state-of-the-art multispectral object detectors. Our code will be released.

Convolutional neural networks (CNNs) have been widely used in change detection of synthetic aperture radar (SAR) images and have been proven to have better precision than traditional methods. A two-stage patch-based deep learning method with a label updating strategy is proposed in this paper. The initial label and mask are generated at the pre-classification stage. Then a two-stage updating strategy is applied to gradually recover changed areas. At the first stage, diversity of training data is gradually restored. The output of the designed CNN network is further processed to generate a new label and a new mask for the following learning iteration. As the diversity of data is ensured after the first stage, pixels within uncertain areas can be easily classified at the second stage. Experiment results on several representative datasets show the effectiveness of our proposed method compared with several existing competitive methods.

Persistent surveillance is one of the major tasks envisioned for unmanned aerial vehicles in that some regions of interest need to be continuously surveyed. This distinction from the one-time coverage/exploration does not allow a straightforward application of the most exploration techniques to address the persistent surveillance problem. This article introduces and demonstrates a method of alterable artificial potential field to control a group of unmanned aerial vehicles to stay over the region of interest. We present attractive potential field to drive the unmanned aerial vehicles to desirable areas, obstacle potential field to move away from obstacles, collision avoidance potential field to control unmanned aerial vehicles not to collide with others, and formation of potential field to make unmanned aerial vehicles gather in a group. The simulation results show that the proposed approach could generate collision-free path for unmanned aerial vehicles staying over the region of interest for a long endurance.

The pose estimation of the aircraft in the taxiing or parking state on the airport surface has been proved helpful for the level of control and command, operation efficiency, capacity of handling special situations of the airports. However, current methods cannot provide satisfied estimate results of the pose of aircrafts because they regard the aircraft as a point. To solve the pose estimation problem of the aircrafts, especially for the small-sized ones, this paper proposes an aircraft pose estimation method based on the contour features. First, the contour features of the aircraft are utilized to design a two-dimensional pose skeleton to show the pose information of the aircraft on the ground. Second, the flux is adopted to represent the two-dimensional aircraft pose skeleton. Finally, a two-branch convolutional neural network is designed to estimate aircraft pose including the aircraft skeleton detection and the aircraft orientation estimation. To overcome the lack of a benchmark dataset, an aircraft dataset was built, and the evaluation of the proposed method's performance on this dataset was also carried out. The experimental results show that our proposed method has satisfied performance compared with other state-of-art approaches on the airport surface aircraft dataset.

The take-off and landing activities of aircraft must operate daily in a safe and orderly manner to provide a safe and efficient transition of passengers and goods. The 6D pose estimation of the aircraft is the key to guaranteeing the safe take-off and landing of the aircraft. The vision-based 6D pose estimation method, an important method when GPS and gyroscope are unavailable, faces the problem of poor estimation accuracy due to large scenes and large depths. An end-to-end 6D pose estimation method of the aircraft is proposed to solve this problem. Firstly, this paper combines the rigid structure characteristic of the aircraft with the direction property of arrows to build an aircraft 3D skeleton with reconstruction ability, simplicity, and direction properties. Secondly, this paper reconstructs the predesigned 3D skeleton of the aircraft from an RGB image and explores the 6D pose information in the reconstructed 3D skeleton. A 3D matrix is used to show the 3D skeleton and improve the encoding of the spatial information in the 3D skeleton. The experimental results show that the proposed method outperforms Wide-Depth-Range by 199% and 105% on the metric ADD and Rete, respectively. Compared with YOLO6D, the proposed method is 58.9% faster.

Low earth orbit (LEO) satellite network can provide services to users anywhere on the earth. However, the high-speed mobility of satellites leads to a dynamic environment, which brings challenges for handover and network performance optimization, especially in the scenario of multiuser using the networks meanwhile. In this paper, we exploit multiple directed graphs to model the handover process for multiuser. The nodes in a graph represent the satellites that the corresponding user may choose to access. The edges represent the possible handovers between adjacent timestamps. The path from the start node to the end node in each graph is the handover strategy of the corresponding user, and the path length is the reward that the user can get. Therefore, the handover strategy problem is transformed into a path planning problem. To minimize the average handover times, maximize the average received power, and minimize the average number of conflicts, we propose a novel handover strategy based on multiobjective multiagent path finding (MOMAPF). The simulated handover experiment on Starlink successfully derives the Pareto-optimal solution set, which corroborates the effectiveness of the proposed handover strategy. The results also show that the proposed strategy has better comprehensive performance than other strategies.