机器人运动的学习控制

机器人学习控制旨在采用简单的控制器结构和易于实现的控制算法,提高机器人轨迹动态跟踪精度。本文讨论了机器人控制研究中存在的困难,简要回顾了早期学习控制思想的形成和发展,重点介绍了近年来国内外有关机器人学习控制研究的现状。     

Multi-mode optimal fuzzy active vibration control of composite beams laminated with photostrictive actuators

Firstly, a multi-field coupling finite element formulation of composited beam laminated with the photostrictive actuators is developed in this paper. Moreover, an optimal fuzzy active control algorithm is also proposed on the basis of the combination of optimal control and fuzzy one.This method opens a new avenue to resolve the contradiction between the linear system control method and nonlinear actuating characteristics of photostrictive actuators. The desired control for suppressing multi-modal vibration of photoelectric laminated beam is firstly obtained through optimal control and then the fuzzy control is used to approach the desired mechanical strain induced by photostrictive actuators. Thus, the multi-mode vibration control of beam is realized.In the design process of optimal fuzzy controller, the design of fuzzy control is independent of optimal control. The simulation results demonstrate that the proposed control method can effectively realize multi-modal vibration control of photoelectric laminated beams, and the control effect of optimal fuzzy control is better than that of optimal state feedback control.     

Application of improved active disturbance rejection control algorithm in tilt quad rotor

The Tilt Quad Rotor (TQR) has complex dynamics characteristics, especially in conversion mode. It is difficult to build the dynamic model of the TQR and the environmental factors have a great influence on it. To solve the problem of control in conversion mode of TQR, this paper carries out the design of the controller based on improved Active Disturbance Rejection Control (ADRC). According to the characteristics of flight in conversion mode, Tracking Differentiator (TD) with explicit model is used to solve the problem of multiple integrals when the system is high-order system. Extended State Observer (ESO) with Radial Basis Function (RBF) neural network is used to estimate and compensate for internal and external uncertainties, and the adaptive sliding mode control in Nonlinear State Error Feedback (NLSEF) is used to improve the response speed of the controller and reduce the parameters which should be tuned. Through the flight control simulation of the TQR, the validity and rationality of the control system are verified.     

黄渤海晴空像元下雾霾识别的机器学习算法

为实时有效地识别和监测雾霾，选取2013年10月4日与10月5日的两景MODIS遥感图像，利用不同类型像元的光谱差异，基于神经网络模型，建立海上雾霾的遥感识别机器学习算法。经检验表明，在晴空条件下，机器学习算法反演精度相对较高，已超过98%，利用算法可以实现对中国近海上空雾霾的监测。     

简论高校实验课程的改革

实验教学是高等院校教学过程的一个重要环节,应建立新型的实验室管理模式,从开放实验室和改革教学方法两个方面出发,阐述了实验课程改革的必要性。     

战斗机嵌入式训练系统中的智能虚拟陪练

智能化"实虚"对抗是现代先进战斗机嵌入式训练系统的重要功能需求。自主空战决策控制技术在未来空战装备发展中扮演关键角色。将当前的功能需求和发展中的技术结合起来,得到了空战智能虚拟陪练的概念。先进控制决策技术的引入使得智能虚拟陪练能够帮助飞行员完成复杂的战术训练,而训练中真实的对抗场景为技术的验证提供了理想的环境,大量的训练数据为技术的持续迭代优化提供了保障。作为可学习和进化的空战战术专家,智能陪练在人机对抗和自我对抗中不断优化,当其具备与人相当甚至超越人的战术能力时,可应用于未来的无人空战系统。智能虚拟陪练需要具备4项基本能力：智能决策能力、知识学习能力、对抗自优化能力和参数化表示能力。对其包含的关键技术进行了分析,提出并实现了一个基于模糊推理、神经网络和强化学习的解决方案,展示了其各项基本能力及目前达到的空战水平。未来更多的模型和算法可在智能虚拟陪练的框架中进行验证和优化。     

Coactive design of explainable agent-based task planning and deep reinforcement learning for human-UAVs teamwork

Unmanned Aerial Vehicles (UAVs) are useful in dangerous and dynamic tasks such as search-and-rescue, forest surveillance, and anti-terrorist operations. These tasks can be solved better through the collaboration of multiple UAVs under human supervision. However, it is still difficult for human to monitor, understand, predict and control the behaviors of the UAVs due to the task complexity as well as the black-box machine learning and planning algorithms being used. In this paper, the coactive design method is adopted to analyze the cognitive capabilities required for the tasks and design the interdependencies among the heterogeneous teammates of UAVs or human for coherent collaboration. Then, an agent-based task planner is proposed to automatically decompose a complex task into a sequence of explainable subtasks under constrains of resources, execution time, social rules and costs. Besides, a deep reinforcement learning approach is designed for the UAVs to learn optimal policies of a flocking behavior and a path planner that are easy for the human operator to understand and control. Finally, a mixed-initiative action selection mechanism is used to evaluate the learned policies as well as the human’s decisions. Experimental results demonstrate the effectiveness of the proposed methods.     

基于高光谱和LiDAR的黄河口湿地植被分类方法

利用无人机（Unmanned Aerial Vehicle, UAV）高光谱影像（Hyper-spectral Imaging, HSI）和激光雷达（Light Detection and Ranging, LiDAR）数据开展黄河口湿地植被分类方法研究。由于高空间分辨率HSI光谱变异性强,以及LiDAR点云密度不均匀,分类结果呈现出“椒盐”现象。为了解决这些问题,本文提出了一种结合空谱特征融合和通道注意力机制的双分支卷积神经网络（SSF-C-DBCNN）。光谱注意力机制通过为每个波段分配不同的权重来减少光谱变异性的影响。空间注意力机制侧重于学习和强调特征表达能力强的密集点云区域空间信息,从而减轻LiDAR点云密度不均匀对结果的影响。最后,在双分支融合特征后引入通道注意力机制来提取更深层次的特征。利用UAV采集的HSI和LiDAR数据进行实验验证,结果表明,本文提出方法的性能优于随机森林和五种深度学习方法,分类结果更为贴合实际土地覆盖,有效地抑制了“椒盐”现象。     

基于学习自动机的雷达干扰资源分配研究

综合考虑干扰机干扰频率、干扰功率、干扰时机和干扰样式等干扰效果评估指标,构建干扰效益矩阵,建立目标函数,构建干扰资源分配博弈模型,并基于学习自动机原理提出了分布式干扰资源分配算法,求解出干扰分配决策。仿真结果表明,提出的分布式干扰资源分配算法能以很低的复杂度来获得很好的干扰性能,有效地解决了干扰资源分配问题,对实际作战指挥有一定的指导意义。     

基于SVM算法的虚假航迹识别

针对云雨杂波和主被动干扰导致多雷达传感器产生虚假目标航迹的问题,利用支持向量机(SVM)算法的自主学习能力,通过构建基于数据驱动的判别模型进行虚假航迹识别。针对航迹起始得到的目标潜在航迹,利用人工智能数据驱动、自学习的特点,设计了SVM算法。通过对已标记真假的目标航迹样本进行离线学习,形成虚假航迹识别的SVM分类器,实现了基于数据驱动的判别模型代替先验知识规则约束的固定模型,并在工程应用中,利用SVM分类器在线识别虚假航迹,完成实时剔除。通过实测雷达数据实验验证,该算法的目标虚假航迹准确率高达95%以上,完全满足实际的工程应用需求。相比基于阈值或规则进行硬性判断的传统虚假航迹识别方法,所提出的算法不仅提高了准确率,还具有较高的实时性,能够适应复杂多变的杂波环境,在实际应用中具有更强的适应性和实用性。因此,提出的基于SVM算法的虚假航迹识别方法对于密集杂波场景下的虚假航迹剔除问题具有显著的实际应用价值。