Transportation system simulation is widely used to support the business processes of transportation agencies including those related to planning, design, operation, management, and safety of increasingly complex multi-modal transportation systems. The advent of connected, automated, shared, and electric (CASE) vehicles provide challenges to be addressed with simulation. This virtual workshop will bring together researchers, vendors, and users of transportation system simulation to identify persistent challenges, discuss solution approaches, present recent research findings, and identify future research needs related to traffic simulation modeling. The discussions will focus on the history, status, and future of traffic simulation, plus modeling of CAVs. The workshop includes invited panels, sessions based on a call for papers/posters, and breakout sessions.
Learn more here.
Title of presentation: Holistic Approach For Simulation And Modelling Of C-ITS Services
Date: Tuesday, November 17, 4:00 pm – 5:30 pm (EST)
Speaker: Tamara Djukic
Narrative: Testing innovative technologies in the domain of traffic with the inclusion of real vehicles and extensive hardware deployment for evaluation of connected vehicle technologies is expensive. Traffic simulation software provides a virtual twin for such infrastructures and equipment, so they are able to provide results for a myriad of cases and situation that would be impossible to achieve in real field, especially at these incipient initial deployments. This is one of main objectives of traffic simulation in C-Roads platform in Europe: understanding the impact of C-ITS services, with its accurate representation in a simulated environment, evaluating multiple hypothetical situations, so giving answers to benefits of a massive roll-out of such services. It is important to recognise the challenge of interoperability and harmonization of C-ITS deployments, and selection of right traffic modelling tools for evaluating the impact of C-ITS services that can provide valuable insights for technology adoption and design.
Objectives: However, a need for more detailed traffic simulation models is becoming even more essential with the arrival of C-ITS services and CAVs, where one of the major applications would be assessment of efficiency and safety. Furthermore, multi-paradigm modeling of CAVs effects requires from traffic simulation models to operate at two extremes of use cases: a) cost-benefit analysis of technology deployment for transport planning and transport management, and b) cost-benefit analysis of new technology where the main aspect is safety. Both use cases have completely opposite requirements, where, for example, use case a) may have strong computational time constraints and case b) may require a high-fidelity representation of the reality. One way to tackle such a complexity of new transport system is to apply holistic approach in development of simulation models.
In this work, we propose a modular simulation framework based on holistic approach that relies on a new architecture of the traffic simulation models.
The main functionalities captured by this framework are:
a) the V2X communication is integrated into the perceptions modules;
b) emulation of external controllers or actuators is provided by original equipment manufacturers;
c) simulates various representation of communication channels, between V2X and V2V (from the physical layer of communication to the application layer);
d) simulates all type of agents involved (pedestrians, bikes, etc); and
e) emulates the new transport management systems, both private or public. The modular simulation framework is implemented in Aimsun Next, where V2X is designed to add communication simulation to a dynamic traffic model.
Methodology: One of the main functions in the new architecture of traffic simulation tool is V2X SDK, designed to allow a modeller to implement connected vehicle VANets within a simulation. A VAnet is an ephemeral network spontaneously created by a collection of connected vehicles in proximity to each other. VANets are formed as vehicles detect transmissions from other vehicles then, while they are in range of each other, they are able to exchange information about their location and activity. By aggregating this information a vehicle is able to infer the pattern of the traffic around it and hence amend its own behaviour accordingly. A VANet may also include a static wireless station in a Roadside Unit, when linked to the network Traffic Management System, it becomes both a data gathering tool and an information dissemination device as a part of the wider C-ITS system.
Design of the proposed holistic approach for traffic simulation is demonstrated for on-going research project C-ROADS Spain. This work presents the impact assessment of three C-ITS services: Road Works Warning, Other Hazardous Notification, and Stationary Vehicle Warning. Services have been tested with different levels equipped vehicles penetration rates with On-board Units, near the installed RSUs, to understand under what conditions there might be an impact – either positive or negative – over traffic conditions. Impact of service has been evaluated with detailed scenario considerations for different level of traffic demand, penetration rates of OBUs and traffic composition to demonstrate impact of service on traffic efficiency and emissions.
Findings and Results: Initial findings confirm expected impacts of the deployment of the Road Works Warning (RWW) service, with an increasing effect as penetration rate is also increased. Globally, the RWW service contributes in achieving a smoother driving conditions, with reduction of unnecessary braking. One interesting aspect is that besides the improvement for informed vehicles, this service also affects non-equipped drivers, as they find a smoother condition nearer the roads works area. In a bit more of detail, improvements in traffic occur at the immediate section (200 m upstream), with better traffic flow conditions, travel times and throughput, and globally. RWW implies certain negative aspects on traffic efficiency around the section where information is received, in this case set at 1000 m upstream, due to the increase of accelerations of decelerations of the equipped vehicles changing lanes and reorganizing themselves. Impacts are sensitive to the location of the road work, as if the road works is downstream of a diverging ramp, even global road conditions could be degraded, with a reduction of 2 km/h in the average speed. In same fashion, these changes in the traffic conditions lead to their corresponding impact in emissions of CO2 and pollutants, improving at global level and in the immediate area of the road works, but with the cost of a slight deterioration upstream, at the section where information is transmitted (at 1000 m of the road works in this specific case).
Conclusions: It is important to recognise the challenge of interoperability and harmonization of C-ITS deployments, and selection of right traffic modelling tools for evaluating the impact of C-ITS services that can provide valuable insights for technology adoption and design. This work proposes modular simulation framework based on holistic approach that relies on a new architecture of the traffic simulation models. The performance and application of the proposed modular simulation framework has been demonstrated for evaluation of C-ITS services, that relies on V2X module that simulates various representation of communication channels and their integration into the human driven vehicle models. Some further steps include collection of real data from pilot (behaviour of equipped vehicles in front of different experiences) to calibrate simulated traffic conditions and reaction of equipped vehicles.
Title of presentation: “Modeling framework of connected and automated vehicles for multi-resolution simulation models” (3941)
Date: Wednesday, November 18, 10:00 am – 11:30 am (EST)
Speaker: Athina Tympakianaki
Co-authors: Leyre Nogues, Jordi Casas, Mark Brackstone
Abstract: During the last decade, there have been numerous studies regarding the development of tools and modeling concepts of connected and autonomous vehicles (CAVs) as well as analyses to understand and quantify their expected implications for society. The modeling of AVs is mainly based on assumptions regarding their technological characteristics and properties, as empirical data are not yet available. Therefore, traffic simulation has been widely used as a means to model the (C)AV characteristics and behavior.
Various simulation-based investigations concerning (C)AVs focus on determining the factors that influence their impacts on road safety, environment, and network efficiency. Most of the studies are conducted in microscopic simulations and their modeling frameworks are based on diverse characteristics of (C)AVs. Narayanan et al. (2020) attempt to determine the common factors that influence the implications of AVs on the road capacity, traffic flow stability and safety. However, the equivalence of the parameters to model their behavior for multi-resolution models (microscopic, mesoscopic, macroscopic) is not sufficiently explored in the literature. Besides the modeling aspects, it is also important to develop methods for the evaluation of the impacts of (C)AVs that can be up-scaled and transferable across different models, network topologies and characteristics, control and traffic patterns. In Friedrich et al. (2019), a modeling framework to integrate AVs into existing macroscopic travel demand models is suggested. However, the suggested approaches regarding the impacts on the network performance and capacity involve extensive analysis on disaggregated network elements (i.e. type of road, intersection). This approach may be impractical and complex for the transferability to other networks, demand and vehicle type compositions.
Current research, as part of the EU-funded project Levitate (https://levitate-project.eu/), aims at providing a range of adequate indicators, derived through simulation, for an evaluation framework to assess the impacts of (C)AVs on the transportation system and society. In line with the project, the objective of this work is two-fold: 1. To propose a generic framework to identify the implications of AVs focusing on the network supply and 2. To integrate the modeling of (C)AVs across multi-resolutions models. Specifically, the first objective relates to the up-scaling and transferability of impacts, independently of the specific modeling assumptions of (C)AVs and network characteristics. The second objective aims to determine the (C)AV modeling parameters and how they should be adjusted in order to reflect the equivalent vehicle behavior and characteristics across different traffic flow resolutions.
To tackle the first objective, a methodology is proposed which utilizes the Macroscopic Fundamental Diagram (MFD) to evaluate the impacts of (C)AVs on the network supply as a functional relationship between the macroscopic network characteristics (i.e. flow, density, and speed). It has been shown that the urban MFD is a property of the network infrastructure itself and independent of the demand patterns (Geroliminis & Daganzo, 2008). Hence, the capacities for (C)AVs can be derived from the urban MFD and for mixed traffic composed by both AV and conventional vehicles (CV) (Kouvelas et al, 2017; Lu et al., 2019). These properties make the MFD a suitable tool which satisfies the transferability and scalability requirements for the impact assessment of AVs.
In particular, macroscopic models use the Volume Delay Functions (VDF) to determine the network travel times. The concept of passenger car unit (PCU) is used to convert the capacity and volumes into passenger car equivalents for each vehicle type. In microscopic models the vehicle characteristics are modeled at a disaggregated level. The simulation outcome can be a characterization of the impacts in terms of macroscopic variables, such as capacity and travel time, from which the VDF can be defined. In mesoscopic models, macroscopic traffic indicators can be also obtained to characterize the impacts of (C)AVs on the network supply. Namely, the capacities and densities derived from the MFD are used as the macroscopic variables in order to investigate the impacts of (C)AVs across microscopic, mesoscopic and macroscopic levels. Moreover, the PCU factors are estimated from the derived capacities and AV penetration rates and can be used to adjust the VDFs for the macroscopic models.
The feasibility and transferability of the proposed methods are explored through simulation experiments for urban networks of various scales, topologies and different traffic flow resolutions. Preliminary results using the MFD demonstrate the positive effects of AVs on the traffic characteristics. In particular, the network capacity and traffic stability increase as the AV penetration rate increases. This can be attributed to the lower reaction times that (C)AVs can achieve due to their enhanced capabilities compared to CV. Following, a regression model is developed for the estimation of the PCU factors for different AV shares based on the corresponding derived capacities. An important outcome of the performed analysis is the consistency in the trend of the impacts of AVs on the network capacities and traffic stability as well as in the estimated PCU factors across the examined networks. Current work continues the determination of the set of parameters to model (C)AVs and the equivalent relationships for the integration to multi-resolution models.