As automated vehicles (AVs) integrate into existing road networks, managing "Transition Areas" (TAs)—locations where automation must be handed back to human drivers—becomes a critical safety challenge. TransAID Deliverable 4.3 details the technical and organizational framework for porting infrastructure-assisted traffic management from research simulations to an integrated, deployment-ready control system (iCS). This transition involves migrating logic from Python-based SUMO environments to a robust C++ architecture within the iTETRIS platform, validated through rigorous aggregate testing and supported by a novel third-party intermediary service model.
Introduction: The Challenge of Transition Areas
Transition Areas represent road segments where environmental complexity or sensor limitations restrict high-level automation. To maintain safety, vehicles must either transition control to the driver or execute minimum-risk maneuvers. Effective management requires real-time coordination between road infrastructure and vehicles. While initial research utilized standalone Python scripts in the SUMO (Simulation of Urban Mobility) environment, these lack the scalability and communication modeling required for real-world deployment.
Technical Implementation: Porting to iCS
The core technical objective was translating traffic management logic into the C++ based integrated control system (iCS).
Modular Architecture
The project utilized the iTETRIS platform, defining a baseApp that manages standardized information exchange between applications and iCS modules. Specific traffic management applications inherit from this baseApp, extending its functionality with procedures defined in previous project phases. This modularity ensures that all TransAID applications share a consistent communication backbone while allowing for use-case-specific logic.
Event-Driven Logic
Unlike time-step simulations, the iCS environment operates on an event-driven model. Applications are designed to handle specific triggers, such as receiving a Cooperative Awareness Message (CAM) or a sensor detection. Key entry points like onAddSubscriptions() and Execute() allow the system to respond dynamically to real-time traffic conditions.
Validation and Testing
To ensure the ported C++ code maintains the original research logic, a testing suite was implemented using the TextTest framework.
Aggregate Metrics
Because exact vehicle trajectories can vary between simulation engines, validation relies on aggregate quantities rather than direct time-space comparisons. Metrics such as mean travel time, network throughput, and mean speed are compared against the original SUMO benchmarks. A small margin of error (epsilon) is permitted to account for the inherent differences in simulation granularity between the two platforms.
The Intermediary Service Model
Beyond technical porting, the paper proposes a "trusted third-party" model to bridge the gap between Original Equipment Manufacturers (OEMs) and road authorities.
Data Privacy and Coordination
OEMs are often reluctant to share raw disengagement data due to proprietary concerns. The intermediary service acts as a buffer, anonymizing and aggregating sensitive data to coordinate management measures. This service matches Infrastructure Support levels for Automated Driving (ISAD) against a vehicle's Operational Design Domain (ODD). Mismatches between infrastructure capabilities and vehicle requirements automatically flag the need for a Transition Area management intervention.
Conclusion and Iterative Development
The successful porting of traffic management measures to the iCS environment marks a significant step toward real-world deployment. By establishing a standardized testing framework and a viable business model for data exchange, TransAID provides a roadmap for managing the complexities of mixed-autonomy traffic.
The iterative nature of this deliverable, spanning from initial drafts in 2019 to the final peer-reviewed version in 2020, reflects the rigorous refinement of both the technical code and the organizational frameworks required for the future of automated transport.
Summary updated
Abstract
This deliverable explains how simulations of both the baseline (WP3) and the traffic management schemes (WP4) can be ported from the SUMO simulation environment (with the help of the TraCI interface and Python scripts) to the iCS environment (using the C++ language).We first gave an explanation on how to set up the creation of a traffic management application in the context of the iCS. Details were given on how to prepare the development of an application, based on the source code in the repository. We also explained the interactions between the iCS, SUMO, ns-3, and the various applications, using subscriptions and the exchange of messages.To this end, the TransAID version of the iTETRIS platform defined in WP6 includes a basic application known as baseAppthat manages the exchange of information between the applications and the iCS modules. The application developed for the different services of the TransAID project will inherit form this baseAppand extend the functionality with the traffic management procedures defined in WP4. In order to develop these applications,a new branch (transaid-apps) is added to the git repository. Note that all TransAID applications developed share the same baseApp. Hence, commits to the baseApp should be strictly separated from thecommits to the TransAID applications in development. Changes to the baseApp as well as other iTETRIS modules like iCS or ns-3 should be integrated into the transaid-dev branch. When porting the traffic management code from the WP4 to the WP6 environments, we need to make sure that the same logic is preserved. In order to guarantee this, all applications implemented in the use cases should create test suites, similarly as described in Deliverable D6.1. We use the same testing framework, called TextTest. Tests are created in the transaid-apps branch of the repository, separated for each use case individually. All tests are stored in the transaid/TransAIDScenarios/tests/scenariosfolder. All relevant data pertaining to a specific use case (i.e. SUMO networks, configuration files, ...) are copied to the relevant scenario in the tests folder. Just as before, the testing concept employed by TextTest is to compare expected output of an entire program run with actual output (output files or stdout and stderr). However, here we need to be a bit more careful and considerate of the complexity involved with comparing the various iCS traffic management applications to their previously created SUMO counterparts.We explain this via a method of aggregate quantities, rather than explicitly comparing time-space diagrams.A more detailed comparison of simulation outputs would be to use detector measurements and/or explicit vehicle trajectories, create time-space diagrams from these (of average speeds or flows), and then compare these with each other and define whether or not the deviation is significant. However, even though this type of analysis would certainly allow us to detect deviations in the time-space plane (e.g.,congested areas that may appear/disappear as artefacts, ...), it would be out of scope. In addition, such analyses have not been done widespread before, as they are also difficult to interpret, and still require some aggregation in order to test these 'automatically'.
BibTeX
@inproceedings{maerivoet2020transaid,
author = {Sven Maerivoet and Kristof Carlier and P. I. Pápics and Robert Alms and Yun-Pang Flötteröd and Leonhard Lücken and Evangelos Mintsis and Vasilios Karagounis and Dimitrios Koutras and Anton Wijbenga and Jaap Vreeswijk and Alejandro Correa and Xiaoyun Zhang and Robbin Blokpoel and Julian Schindler},
title = {{TransAID Deliverable 4.3 (second iteration): Translation of traffic management measures to iCS, scale-up, and wider deployment}},
booktitle = {Elib},
year = {2020},
}