Parking trajectory characteristics were analyzed, and an approach for trajectory planning was proposed on the basis of the B spline theory, and the parking path was realized by adjusting the control points. Potential collision points were analyzed,and the constraint equations were established accordingly. The orientation constraints at the start point,the location constraints at the parking point,and the restriction of vehicle were considered, and the constraint equations were established respectively. Parking kinematic function was presented on the basis of Ackerman steering theory, and parking trajectory optimization function was presented to minimize the max trajectory curvature in parking process with the restrictions mentioned above. The approach was used to plan parking path for general parking space and tight parking space, and the trajectory function was solved in Matlab. The simulation results have shown that the car can reverse into parking set safely in general parking space, with the trajectory curvature changing continuously and the car paralleling the parking set at the parking point. The car can reverse into parking set without collision in tight parking space by adjusting steering angle repeatedly, with the trajectory curvature changing continuously. It is proved that the approach proposed can find a collision-free path in different parking spaces and meets the demand of continuous trajectory curvature, so the problem of shutting down to steer the wheel in parking process has been solved.