A5017059438- Adaptive Control of Nonlinear Systems
- Guidance and Control Systems
- Distributed Control Multi-Agent Systems
- UAV Applications and Optimization
- Remote Sensing in Agriculture
- Video Surveillance and Tracking Methods
- Military Defense Systems Analysis
- 3D Shape Modeling and Analysis
- Robotic Path Planning Algorithms
- Infrared Target Detection Methodologies
- Robotics and Sensor-Based Localization
- Neural Networks Stability and Synchronization
- Advanced Wireless Communication Technologies
- Advanced Vision and Imaging
- Opportunistic and Delay-Tolerant Networks
- Target Tracking and Data Fusion in Sensor Networks
- Species Distribution and Climate Change
- Image Processing and 3D Reconstruction
- Electromagnetic Launch and Propulsion Technology
- Computational Fluid Dynamics and Aerodynamics
- Impact of Light on Environment and Health
National University of Defense Technology
2017-2024
The Visual Object Tracking challenge VOT2017 is the fifth annual tracker benchmarking activity organized by VOT initiative. Results of 51 trackers are presented; many state-of-the-art published at major computer vision conferences or journals in recent years. evaluation included standard and other popular methodologies a new "real-time" experiment simulating situation where processes images as if provided continuously running sensor. Performance tested typically far exceeds baselines. source...
Clusters of unmanned aerial vehicles (UAVs) are often used to perform complex tasks. In such clusters, the reliability communication network connecting UAVs is an essential factor in their collective efficiency. Due wireless environment, however, malfunctions within cluster likely during flight UAVs. cases, it important control and rebuild connected network. The asymmetry topology also increases complexity mechanisms. traditional methods based on consistency rely motion information...
Distributed control method plays an important role in the formation of a multi-agent system (MAS), which is prerequisite for MAS to complete its missions. However, lack considering collision risk between agents makes many distributed methods lose practicability. In this article, that takes avoidance into account proposed. At first, problem can be divided pair-wise unit problems where each agent moves expected position and only needs avoid one obstacle. Then, deep Q network (DQN) applied...
A class of arbitrary time convergence controllers based on a special time-varying scaling function provide designers with good choice to realize prescribed stable systems. However, these have the problem conservative control parameter range and lack uniform formula for systems different order. Herein, controller is improved, unified <a:math xmlns:a="http://www.w3.org/1998/Math/MathML" id="M2"> <a:mi>n</a:mi> </a:math> -th order system given, more accurate obtained. By constructing an...
Structural SVM trackers and correlation filter have demonstrated dominant performance in recent object tracking benchmarks. However, structural naturally suffer from shortage of samples low speed, time-consuming adaption is need to relieve the boundary effects. Thus, we design a jointed tracker by concatenating high-speed SSVM method-DSLT multi feature CF method-STAPLE realize advantage complementation. We show that precision robustness can be improve large margin comparing either single...
The design of guidance law can be considered a kind finite-time error-tracking problem. A unified free-time convergent approach based on the error dynamics and convergence method is proposed in this paper. Firstly, desired proposed, its time set freely, which independent initial states parameters. Then, illustrative laws considering leading angle constraint, impact constraint are derived respectively. connection distinction between existing analyzed theoretically. Finally, performance...
Three-dimensional tracking of multiple objects from views has a wide range applications, especially in the study bio-cluster behavior which requires precise trajectories research objects. However, there are significant temporal-spatial association uncertainties when similar to each other, frequently maneuver, and cluster large numbers. Aiming at such multi-view multi-object 3D scenario, current statistical model based Kalman particle filter (CSKPF) method is proposed following Bayesian...
Because the scanning point cloud of packing box is a bottom-free cuboid shape, traditional Iterative Closest Point (ICP) algorithm cannot realize registration and CAD model, thus error detection shape. To solve this problem, paper presents incomplete with shape model. Firstly, method quaternion-based surface normal vector alignment proposed for initial registration. Then, to achieve fine registration, new Attribute Surface Labelling (ASL) establish nearest neighbor set improve ICP algorithm....