Robust Nonlinear Adaptive Control of Unmanned Aerial Vehicles Using Neural Networks

使用神经网络对无人机进行鲁棒非线性自适应控制

• 批准号: 1102382
• 负责人: Subodh Bhandari
• 金额: $ 36万
• 依托单位: Cal Poly Pomona Foundation, Inc.
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2011
• 资助国家: 美国
• 起止时间: 2011-08-15 至 2016-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1102382&HistoricalAwards=false
• 关键词: Robust; Nonlinear; Adaptive; Control; Unmanned

The objective of this research is to develop and experimentally validate robust nonlinear adaptive controllers for unmanned aerial vehicles (UAVs) using neural networks. The approach is to use both pre-trained and on-line neural networks based on multilayer perceptrons. The concept of dynamic nonlinear damping will be applied to modify the adaptation law to increase the robustness in the presence of unmodeled dynamics. A stable weight adjustment rule for the on-line neural network will be derived using a Lyapunov stability theorem. Intellectual MeritThis research, through the use of adaptive controllers, will increase the robustness of UAV operations. The lack of robustness has limited the use of UAVs, despite their potential to replace manned vehicles for many types of missions. Also, as the level of UAV autonomy increases, it is important to design controllers that can work in the entire flight envelope. This research will involve an interdisciplinary team of faculty members from the Engineering and Computer Science Departments at Cal Poly Pomona. An existing fleet of UAVs and complimentary resources will be used. Broader ImpactIncreased UAV robustness will have potential to extend and increase the use of UAVs for rescue missions, border patrol, scientific research, etc., which will have a significant positive impact on society. This research will involve a number of undergraduate and underrepresented students. This research will also be helpful in curriculum improvement. Moreover, this project will help enhance the research capability of the University, a Hispanic Serving and Designated Undergraduate Institution. Research results will be disseminated to the Intelligent Systems community via conference presentations and publication in journals.

本研究的目的是开发和实验验证鲁棒非线性自适应控制器的无人机（UAV）使用神经网络。该方法是使用预先训练和在线神经网络的多层感知器的基础上。动态非线性阻尼的概念将被应用到修改的自适应律，以增加鲁棒性的存在未建模动态。一个稳定的权值调整规则的在线神经网络将使用李雅普诺夫稳定性定理。智能优点这项研究，通过使用自适应控制器，将增加无人机操作的鲁棒性。尽管无人机有可能取代载人飞行器执行许多类型的任务，但缺乏鲁棒性限制了无人机的使用。此外，随着无人机自主水平的提高，设计能够在整个飞行包线内工作的控制器也很重要。这项研究将涉及来自加州理工大学波莫纳分校工程和计算机科学系的跨学科教师团队。将使用现有的无人机机队和免费资源。更广泛的影响增强的无人机鲁棒性将有可能扩展和增加无人机在救援任务、边境巡逻、科学研究等方面的使用，这将对社会产生重大的积极影响。这项研究将涉及一些本科生和代表性不足的学生。本研究也将有助于课程的改进。此外，这个项目将有助于提高大学的研究能力，西班牙裔服务和指定的本科院校。研究成果将通过会议演示和期刊出版传播到智能系统社区。