In order to study the stability of vehicle active front wheel steering (AFS) system in complex environment, an active front wheel steering active disturbance rejection control (ADRC) method based on RBF neural network is proposed. By designing the structure of RBF neural network, the gradient descent method is adopted to achieve the purpose of parameter tuning of active front wheel steering system (AFS); Aiming at the deficiency of fixed transmission ratio of AFS, the ideal variable transmission ratio rule with fixed yaw rate gain is designed. The results show that the ADRC strategy based on RBF neural network can effectively track the ideal yaw rate of active front wheel steering. Compared with ADRC, the anti-interference amount is increased by 25.8%, and the oscillation amplitude of yaw rate is effectively suppressed. It can be seen that this method improves the tracking ability of ideal yaw rate, improves the controllability and stability of the vehicle, and has good control accuracy.