Abstract This paper presents three fault-tolerant control (FTC) strategies for a coaxial octorotor unmanned aerial vehicle (UAV) regarding motor failures. The first FTC is based on a control mixing strategy which consists of a set of control laws designed offline, each one dedicated to a specific fault situation. The second FTC, a robust adaptive sliding mode control allocation is presented, where the control gains of the controller are adjusted online in order to redistribute the control signals among the healthy motors in order to stabilize the overall system. The third FTC strategy is a new strategy proposed in this article, which is based on a self-tuning sliding mode control (STSMC) where the control gains are readjusted based on the detected error to maintain the stability of the system. Multiple indoor experiments on an octorotor UAV are conducted to show and compare the effectiveness and the behavior of each FTC scheme after successive faults are injected. More specifically, we inject complete actuator’s failures into the top four motors of our octorotor. Every strategies show good fault tolerance results, although the control mixing method performs slightly better overall while the adaptive method performs slightly worse. However, the control mixing method requires a huge design effort to take into account as much situations as possible, while the adaptive method and the STSMC only require to determine a few gains. The adaptive method do not need fault detection to operate, but it thus does not provide information on the system’s health without an additional fault identification and diagnosis mechanism, while both the control mixing method and the STSMC provide such information.

中文翻译：

---

八旋翼无人机连续故障自调谐、自适应和多路复用FTC策略的比较研究

摘要 本文针对同轴八旋翼无人机 (UAV) 针对电机故障提出了三种容错控制 (FTC) 策略。第一个 FTC 基于控制混合策略，该策略由一组离线设计的控制律组成，每个控制律专用于特定的故障情况。第二个 FTC，一种稳健的自适应滑模控制分配，其中控制器的控制增益在线调整，以便在健康电机之间重新分配控制信号，从而稳定整个系统。第三种 FTC 策略是本文提出的一种新策略，它基于自调整滑模控制 (STSMC)，其中根据检测到的误差重新调整控制增益，以保持系统的稳定性。对八旋翼无人机进行了多次室内实验，以展示和比较每个 FTC 方案在注入连续故障后的有效性和行为。更具体地说，我们将执行器的完整故障注入我们八旋翼机的前四个电机中。每种策略都显示出良好的容错结果，尽管控制混合方法总体上表现稍好，而自适应方法表现稍差。然而，控制混合方法需要大量的设计工作来考虑尽可能多的情况，而自适应方法和 STSMC 只需要确定几个增益。自适应方法不需要故障检测来运行，但如果没有额外的故障识别和诊断机制，它不会提供系统健康的信息，