Development and Validation of an Open Architecture for Autonomous Vehicle Control

Abstract

Teams dedicated to research in the field of autonomous vehicles can be found in many universities and research centers, but they often face the challenge of finding a suitable platform for their work. The primary reason for this challenge is the inaccessibility and high cost of commercial autonomous vehicles, leading researchers to rely on simulators.This paper introduces a new software architecture designed to automate vehicles, providing all the necessary capabilities of an autonomous vehicle in a more cost-effective and efficient manner. The architecture is designed to be modular, universal, and with a public interface, making it easy to modify, adapt to any type of vehicle, and accessible to any researcher.The new software architecture has been implemented in two platforms: a vehicle integrated with OpenPilot via ROS2 without any external hardware, and a last-mile robot. A validation test was conducted with volunteers to assess the reaction of passengers while the car was driving autonomously. The results of this implementation demonstrate the potential of this new software architecture to provide a comprehensive and accessible platform for the advancement of autonomous vehicle research.