In order to improve the safety, parking efficiency and smoothness of the parallel parking of driverless cars, for the parking process needs to meet the path curve curvature is small, parking distance is short, the lane occupation is less, through the analysis of the characteristics of the parallel parking working conditions, put forward a NSGA-II algorithm based on the parking starting point of multi-objective optimization method. Firstly, based on the analysis of vehicle kinematics model, the quintic polynomial curve is used to construct the parking path, and the analytical function between the parking start point and various indexes is established; then, the NSGA-II algorithm is used to optimize the constructed multi-objective model, so as to obtain the optimal optional set of the parking start point; finally, the results of the simulation experiments show that: in order to satisfy the various objectives of the synergistic optimization as well as constraints, the proposed method of the design of the parking start point can achieve the parking path curve of less curvature and shorter parking distance.