Author(s)

SARA GHATTA, SORAYA ZENHARI, AMIRHASSAN MONADJEMI

Abstract

In today’s busy world, parking a car in crowded cities is time-consuming and challenging. Parallel parking systems are a valuable innovation, especially for disabled and inexperienced drivers, as they can prevent collisions. Despite their common use, artificial neural networks (ANNs) and backpropagation algorithms (BP) have issues, such as difficulty in estimating network configurations and low accuracy. While BP is traditionally used to train ANNs, it is not effective for parking a car. To achieve accurate vehicle control, a self-organising map (SOM) clusters data from various parking scenarios and uses this information to classify the data. According to simulations conducted using MATLAB, SOM provides better accuracy and reliability for parking cars compared to BP.

Keywords

self-organising map algorithm, automated vehicle, car-like robot parallel parking, artificial neural network (ANN), robot navigation, data scaling