Abstract: To improve the anti-disturbance capability of controller in automatic parking system and optimize the execution effect of control law, the closed-loop layered controller of front wheel angle was designed. The non-smooth control law based on the fal function was designed with driving distance as non-time reference to output the target front wheel angle. The front wheel angle observer was designed based on the Ackerman steering model, and the fuzzy sliding mode controller was used to implement closed-loop lateral control. The Carsim/Simulink co-simulation was built to verify the feasibility, tracking effect and robustness of the designed controller by simulating typical parking scenarios. A real vehicle test platform was used for testing. The results show that the closed-loop layered controller of front wheel angle can quickly and accurately track the target path, improve the lateral control accuracy of the parking system and ensure the tracking control effect under unknown steering nonlinear disturbances.