A Tutorial on Real-time Convex Optimization Based Guidance and Control for Aerospace Applications
A Tutorial on Real-time Convex Optimization Based Guidance and Control for Aerospace Applications
复制标题
基于实时凸优化的航空航天应用制导与控制教程
DOI：
10.23919/acc.2018.8430984
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
Behçet Açikmese
中科院分区：
文献类型：
--
作者：
Y. Mao;Michael Szmuk;Behçet Açikmese
具有挑战性的控制问题在航空航天工程应用中无处不在。这些应用包括可重复使用的火箭、航天器交会和对接、用于地面成像的卫星星座管理，以及许多其他要求车辆在遵守物理和任务限制的情况下可靠运行的应用。物理限制是由于有限的燃料等限制，而任务限制可能来自传感器指向要求或安全保护区域。通常，任务的成功需要使用有限的机载计算资源实时满足这些约束条件。在本文中，我们提供了如何在基于优化的控制框架中制定一些航空航天控制问题实例的教程。具体来说，我们将控制问题分解为两个层次：制导（轨迹优化）和反馈控制。在指导中，我们利用最近的凸化结果将非凸轨迹优化问题表述为有限维凸优化问题，并使用快速可靠的内点法（IPM）算法求解。我们讨论了当前凸化技术的现状，并概述了这些技术应该使用的背景。最后，我们概述了反馈控制律的综合方法。这些控制律用于在给定误差范围内，以及存在模型不确定性和环境干扰的情况下跟踪制导轨迹。
Challenging control problems are ubiquitous in aerospace engineering applications. Such applications include reusable rockets, spacecraft rendezvous and docking, satellite constellation management for terrestrial imaging, and many other ones where vehicles are required to perform reliably while adhering to physical and mission constraints. Physical constraints are due to limitations like finite fuel, whereas mission constraints can arise from sensor pointing requirements or safety keep-out zones. Often, mission success necessitates that these constraints be satisfied in real-time, using limited on-board computational resources. In this paper, we present a tutorial on how to formulate some aerospace control problem examples in an optimization based control framework. Specifically, we decompose the control problem into two levels: guidance (trajectory optimization), and feedback control. In guidance, we use recent convexification results to formulate non-convex trajectory optimization problems as finite-dimensional convex optimization problems, which we solve using fast and reliable Interior Point Method (IPM) algorithms. We discuss the current state of the art of convexification techniques, and outline the context in which these techniques should be used. Lastly, we present an overview of synthesis methods for feedback control laws. These control laws are used to track the guidance trajectories within specified error bounds, and in the presence of model uncertainties and environmental disturbances.