Machine Intelligence Flight Control mimicking Test Pilot Function

自动飞行控制技术的进步使开发全自动无人机成为可能。然而，在实际开发中，在试飞过程中因碰撞而损坏车辆的风险并不小，因此，需要像人类试飞员那样实现更可靠的飞行控制。为了解决这一问题，它追求两大功能，即实时监测飞行控制的安全裕度和利用试飞数据实时调整飞行控制参数。为了实现这些功能，提出了一种新的飞行控制律设计方法--层次结构动态逆方法。通过考虑各种不确定性的真实飞行仿真分析，对该方法进行了评估。取得了令人振奋的结果。

Progress in automatic flight control technology has made it possible to develop a fully automatic unmanned aerial vehicle. In real development, however, risk of vehicle damage by crash through the flight test is not small, therefore, more reliable flight control like what is realized by a human test pilot is desirable. In order to solve this problem, it pursued two major functions, that is, real time monitoring of safety margin of the flight control and real time adjustment of flight control parameters by using flight test data. A novel flight control law design approach called Hierarch-structured Dynamic Inversion was proposed to realize these functions. The approach was evaluated by a realistic flight simulation analysis which considered various kinds of uncertainty. It obtained promising results.