An Automatic Parking Algorithm Design Using Multi-Objective Particle Swarm Optimization

In this article, an automatic vehicle parallel parking algorithm, consisting of path planning, controller design, and state estimation is developed. The path is planned using clothoid sequences and a straight line, which avoids stopping the car to reorient the wheels. The control inputs, including speed and steering angle, are a function of traveled distance. This method enables the car to park from different initial poses, achieving reduced parking time and the ability to park in one or two maneuvers, in smaller than standard places. An evolutionary optimization algorithm is used to calculate the best speed parameter according to the defined criteria. The proposed technique utilizes the Unscented Kalman Filter (UKF) to estimate the traveled distance, resulting in a smaller error compared to the conventional Extended Kalman Filter (EKF). The research aims to introduce an optimal automatic parking algorithm to improve the existing methods in terms of parking duration, the required space size for parking in the maximum of two maneuvers, and path continuity. Finally, the fidelity and improved performance of the proposed method are assessed in various probable conditions using the powerful Monte Carlo simulations.

Automated summary

This article develops an automatic vehicle parallel parking algorithm that includes path planning, controller design, and state estimation. The path is planned using clothoid sequences and a straight line, without the need for stopping the car to reorient the wheels. The control inputs are a function of traveled distance. The proposed technique aims to improve existing methods in terms of parking duration, space requirements, and path continuity, and it is evaluated using Monte Carlo simulations.