This paper presents the development of an intelligent system for the detection of parking spaces in 1:10 scale autonomous vehicles, using HC-SR04 ultrasonic sensors and machine learning techniques. Distance data were collected using an ESP32 microcontroller, allowing continuous transmission at a frequency of 250 Hz. This information was structured in windows of 60 samples and processed through metrics of central tendency, dispersion and distribution shape. Two supervised classification models were trained: multiple logistic regression and support vector machines (SVM) with an RBF kernel. Both models showed good predictive capacity, with the SVM model standing out for its slight superiority in precision and accuracy. The results validate the feasibility of using low-cost sensors and machine learning models in solving perception tasks for autonomous vehicles at scale. This proposal constitutes a solid basis for the development of real-time autonomous parking systems with low resource consumption, applicable in educational environments or functional prototypes.