航天器
容错
旋转(数学)
计算机科学
姿态控制
控制理论(社会学)
航空航天工程
控制(管理)
物理
分布式计算
工程类
人工智能
作者
David van Wijk,Manoranjan Majji,Kerianne L. Hobbs
摘要
This paper investigates the problem of safety-critical control in the presence of faults and rapidly variable dynamics using sequential estimation and run time assurance. Safety filtering or run time assurance (RTA) approaches can be used to monitor the desired control of a primary controller and prevent the system from taking unsafe actions by satisfying safety constraints online. However, these methods often assume known dynamics, or at the very least, constant dynamics during operation. In the case of some faults such as a collision, this assumption may degrade. Thus, a constrained multiplicative extended Kalman filter (CMEKF) is developed to estimate the dynamics in tandem with a fault detection scheme and is used to mitigate errors induced by faults. The CMEKF informs the control barrier functions (CBFs) used for safety assurance. The framework is validated through an attitude control problem on a simulated rigid body spacecraft and it is shown that this framework sufficiently assures safety in the presence of abrupt state change faults.
科研通智能强力驱动
Strongly Powered by AbleSci AI