A real-time embedded drive-by-wire control module for self-driving cars with ROS2

Abstract

The objective for attaining safety and reliability for an autonomous driving vehicle has provided numerous challenges for researchers. The widely used robot operating system (ROS) is centralised and wastes computing resources due to innate features of the architecture. This paper proposes a safety-critical Simulink compatible real-time operating system (RTOS)-based high-low level controller framework with real-time online parameter tuning for self-driving cars. In addition, a method for leveraging the new features of robot operating system 2 (ROS2), an enhanced safe and reliable self-driving platform, was designed and built. Two geometric path trackers, namely the pure pursuit controller and the Stanley controller were implemented on this high-low level controller framework. Both controllers were evaluated using both indoor and outdoor experiments with reliable performance. This is the first application of ROS2 on a small-scale Ackermann steering platform, and in the use of Simulink modelling for rapid control system prototyping.