Abstract: This paper proposes a redundant strategy for actuator failures in autonomous vehicles, including steering and braking actuator fault. Firstly, the unscented Kalman filter is used to estimate the state of the steering angle input of the command layer, and judge whether the actuator is working normally according to the deviation of the estimated state and the actual state. After determining the steering failure, the direct yaw moment control is adopted to perform the redundant steering by ESC. In addition, after the brake failure is determined, the desired pressure is obtained through the MAP method and ESC is used to achieve braking redundancy. The simulation test results show that actuator faults can be diagnosed timely and effectively, and corresponding functional redundancy can be realized to ensure vehicle driving safety.

Key words: Autonomous vehicle; Steering actuator; Braking actuator; Fault diagnosis; Fault tolerance

摘要： 提出一种自动驾驶汽车执行器故障冗余策略，包括转向和制动执行器故障。首先，利 用无迹卡尔曼对命令层的转角输入进行状态估计，并根据估计状态量与实际状态量的偏差判 断执行器是否工作正常。在确定转向故障后，利用直接横摆力矩控制依托汽车电子稳定控制 系统进行冗余转向控制；而确定制动故障后，通过 MAP 图查询法进行制动压力的获取，并借 助汽车电子稳定控制系统实现制动执行器的冗余。仿真结果表明，执行器故障能得到及时有 效的诊断，并能获得相应的功能冗余，保证紧急工况下车辆安全。

关键词: 自动驾驶汽车；转向执行器；制动执行器；故障检测；故障容错