Assembly of a test bench for road sign recognition and decision-making software for autonomous vehicles

Abstract

This project develops a test bench and a decision-making system aimed at road sign recognition and decision-making for vehicles. Using a Monte Carlo simulation method to generate data, the system analyzes and responds to inputs such as traffic signs and pedal commands; the system features an RFID sensor and a computer vision system. Communication occurs through a developed serial protocol between an ATmega328P microcontroller and a Raspberry Pi, which facilitates sending information to the display, while a moving average filter improves the accuracy of the potentiometer data readings. The tests performed showed efficiency in sign recognition and stability in decision-making. The test bench allows the system to execute actions according to a pre-defined priority level, where pedestrian and stop signs have priority to ensure safety. The results show that covering more conditions in the decision system can increase its effectiveness in real-world environments. It is concluded that the solution is a promising platform for research in autonomous technology and offers potential for future enhancements, including diverse controls, integration with automotive networks, and improvements in communication with sensors and connected vehicle systems