This paper focuses on the fault-tolerant control of unmanned vehicles with sensor and actuator faults, which proposes an observer-based active composite hierarchical fault-tolerant controller. By dividing the total faults into sensor and actuator faults, an extended state observer based on neural network is used to handle actuator fault, and sensor fault is estimated through the disturbance observer. Under the framework of backstepping control, following the Lyapunov stability theory, the boundedness of the closed loop system is proved. Finally, simulations and experiments are conducted using breeding unmanned systems to verify its effectiveness, including breeding ground unmanned vehicles and quadrotor unmanned aerial vehicle.