Aerial object detection acts a pivotal role in searching and tracking applications. However, the large model, limited memory, and computing power of embedded devices restrict aerial pedestrian detection algorithms’ deployment on the UAV (unmanned aerial vehicle) platform. In this paper, an innovative method of aerial infrared YOLO (AIR-YOLOv3) is proposed, which combines network pruning and the YOLOv3 method. Firstly, to achieve a more appropriate number and size of the prior boxes, the prior boxes are reclustered. Then, to accelerate the inference speed on the premise of ensuring the detection accuracy, we introduced Smooth-L1 regularization on channel scale factors, and we pruned the channels and layers with less feature information to obtain a pruned YOLOv3 model. Meanwhile, we proposed the self-built aerial infrared dataset and designed ablation experiments to perform model evaluation well. Experimental results show that the AP (average precision) of AIR-YOLOv3 is 91.5% and the model size is 10.7 MB (megabyte). Compared to the original YOLOv3, its model volume compressed by 228.7 MB, nearly 95.5 %, while the model AP decreased by only 1.7%. The calculation amount is reduced by about 2/3, and the inference speed on the airborne TX2 has been increased from 3.7 FPS (frames per second) to 8 FPS.

Visual-based object detection and understanding is an important problem in computer vision and signal processing. Due to their advantages of high mobility and easy deployment, unmanned aerial vehicles (UAV) have become a flexible monitoring platform in recent years. However, visible-light-based methods are often greatly influenced by the environment. As a result, a single type of feature derived from aerial monitoring videos is often insufficient to characterize variations among different abnormal crowd behaviors. To address this, we propose combining two types of features to better represent behavior, namely, multitask cascading CNN (MC-CNN) and multiscale infrared optical flow (MIR-OF), capturing both crowd density and average speed and the appearances of the crowd behaviors, respectively. First, an infrared (IR) camera and Nvidia Jetson TX1 were chosen as an infrared vision system. Since there are no published infrared-based aerial abnormal-behavior datasets, we provide a new infrared aerial dataset named the IR-flying dataset, which includes sample pictures and videos in different scenes of public areas. Second, MC-CNN was used to estimate the crowd density. Third, MIR-OF was designed to characterize the average speed of crowd. Finally, considering two typical abnormal crowd behaviors of crowd aggregating and crowd escaping, the experimental results show that the monitoring UAV system can detect abnormal crowd behaviors in public areas effectively.

As a very serious problem for control systems, actuator saturation often leads to an unstable system. The maneuvering flight of quadrotors may lead to actuator saturation, which commonly affects the flight quality, including trajectory-tracking accuracy. Quadrotors usually adopt double-closed-loop control, of which the outer loop is the position control used to calculate the desired attitude angle and the inner loop is the attitude control used to generate control actions for the vehicle. To address actuator saturation during the maneuvering flight of quadrotors, an attitude controller based on the conditioned super-twisting algorithm (CSTA) was designed. The sign functions in CSTA were replaced by a hyperbolic tangent function, which suppresses the chattering of CSTA. In order to improve the rapidity of the improved controller, the tanh-CSTA (TCSTA), a gain adjustment factor was utilized. The rapidity and smoothness of the controller can be satisfied by adjusting the gain factor. Additionally, a sliding surface was designed according to the characteristics of the quadrotor control structure, so that the TCSTA controller can be applied to the second-order system. The performances of the designed improved controller were simulated using MATLAB, and the results showed that the designed controller is more robust than the PID controller. Moreover, compared with the CSTA controller, the chattering of the designed controller was obviously suppressed without reducing the control accuracy.

To solve the problem with complex structure and low resource utilization of traditional target tracking UAV system based on visual guidance, this paper discusses an implementation method of quadrotor UAV target tracking system based on OpenMV. This system mainly includes flight control system and visual guidance system. TM4C123GH6PMI was selected as the control unit for flight control system. Six-axis motion sensor and digital magnetometer were combined into the pose measurement module. And ultrasonic sensor was selected to get the altitude information for drone. This system took OpenMV as the visual guidance system for target detection and position. For the real-time and accuracy requirements of the tracking system, the centroid-based tracking algorithm was used for calculating and resolving the target. The experiment results show that the system is stable and reliable.

Multi-view clustering algorithms have been successfully used in different consumer electronic products, such as common digital cameras and unmanned vehicles. Currently, existing multi-view graph clustering (MVGC) methods learn the similarity of directly connected samples for clustering. However, these MVGC methods can not fully consider the indirect relation among samples and high-order relation across multi-view data. In this paper, a new multi-view comprehensive graph clustering (MCGC) method is devised, which can fully learn the similarity based on (1) first-order proximity (FOP) (i.e. the direct relation of pairwise samples); (2) second-order proximity (SOP) (i.e. the indirect relation of pairwise samples); and (3) third-order proximity (TOP) (i.e. the three-order relation of multiple views). Since the operations of these three components are iteratively carried out, the interaction between similarity learning can be encouraged and the comprehensive graph can be generated effectively for clustering. In-depth experiments on six commonly benchmark datasets show the superiority of the MCGC method.