Core Concepts
The authors propose CARLOS, an open, modular, and scalable simulation framework that leverages the CARLA and ROS ecosystems to enable efficient development and testing of software for cooperative intelligent transport systems (C-ITS).
Abstract
The paper presents a novel simulation architecture and its implementation in the CARLOS framework. The key points are:
Use case analysis: The authors identify three main use cases for simulations in the context of automated driving - software prototyping, data-driven development, and automated testing. These use cases have varying requirements in terms of scenario complexity, simulation fidelity, and scalability.
Architecture design: The proposed architecture follows a microservice approach, dividing the simulation into modular components across different layers (simulation, storage, application). This enables flexibility, maintainability, and scalability through containerization and orchestration.
CARLOS implementation: The authors implement the proposed architecture in the open-source CARLOS framework, which builds upon the CARLA simulator and the ROS ecosystem. CARLOS provides core building blocks like a ROS bridge, scenario runner, and a data generation pipeline.
Evaluation: The authors evaluate CARLOS against the native CARLA ecosystem, highlighting improvements in usability, maintainability, interoperability, scalability, and test capabilities. They also demonstrate the integration of a custom perception module within the CARLOS framework.
The CARLOS framework aims to provide an open, flexible, and scalable foundation for the development and testing of C-ITS software, leveraging the strengths of the CARLA and ROS communities.