Robust State and Uncertainty Estimation for Unmanned Systems in the Presence of External Uncertainties

该奖项的研究目标是开发一个框架，用于估计运动状态和不确定的外部输入，用于在未知环境中运行的无人驾驶车辆的反馈控制。该框架将专门解决由于外部输入不确定、运动传感器中的传感器噪声以及反馈控制中的时延而引起的不确定性。对车辆动力学的全局处理将使在存在不确定性的情况下估计和控制快速机动的车辆的运动成为可能。估计方案是确定性的，但近似描述车辆动力学中的不确定性的潜在概率密度函数。在存在有界时滞的情况下，利用状态估计和不确定输入的鲁棒控制方案来稳定和控制车辆。一种基于力学离散时间解释的计算算法将被用于实时实施估计和控制方案。可交付内容包括此框架组件的软件实现、研究结果文档和工程专业学生教育。拟议的研究将直接影响无人系统在鲜为人知的环境中的应用。示例应用包括空间探索、卫星大地测量、热带风暴数据收集、水下管道泄漏检测以及搜索和救援任务。通过这项研究实现的软件将改善用于此类应用的无人驾驶系统的可靠性。特别是，这项研究的结果将(1)改善对影响无人驾驶车辆运动的不确定输入的表征，(2)使无人驾驶车辆能够学习和适应其环境中的不确定性，(3)表征和利用运动控制中的时延。工程专业的研究生和本科生将从课堂教学和参与研究中受益。将开发一个教育网站，向学生、专家和广大社区传播研究成果。

The research objective of this award is to develop a framework for estimation of motion states and uncertain external inputs for feedback control of unmanned vehicles operating in poorly known environments. This framework will specifically address uncertainties due to uncertain external inputs, sensor noise in motion sensors, and time delay in feedback control. A global treatment of vehicle dynamics will make it possible to estimate and control the motion of rapidly maneuvering vehicles in the presence of uncertainties. The estimation schemes are deterministic, but approximate the underlying probability density function describing the uncertainty in the dynamics of the vehicle. A robust control scheme using the estimates of states and uncertain inputs in the presence of bounded time-delay will be formulated to stabilize and control the vehicle. A computational algorithm, based on a discrete-time interpretation of mechanics, will be used to implement the estimation and control schemes in real-time. Deliverables include software implementations of the components of this framework, documentation of research results, and engineering student education. The proposed research will directly impact applications of unmanned systems in poorly known environments. Example applications include space exploration, satellite geodesy, data gathering from tropical storms, underwater pipeline leak detection, and search and rescue missions. Software made possible through this research will offer improvements to the reliability of unmanned systems used in such applications. In particular, results from this research will (1) improve characterization of uncertain inputs affecting the motion of unmanned vehicles, (2) enable the unmanned vehicle to learn and adapt to the uncertainties in its environment, and (3) characterize and exploit time delay in motion control. Graduate and undergraduate engineering students will benefit through classroom instruction and involvement in the research. An educational website will be developed to disseminate research results to students, experts, and the community at large.