You are hardly ever alone in the world. How, where and when other participants in a simulated environment move is subject to a traffic simulation component. The behavior of participants may be scripted or intelligent, they might be wheel-based, on their own feet or in the air. Infrastructure influencing the flow of entities or communicating with them may be part of a traffic component.
It might sometimes be hard to distinguish between traffic and scenario simulation but we consider traffic simulation the core execution component of scenario simulation which acts on a definite set of parameters and focuses on participants only (i.e. neither weather nor other environment features).
The rating criteria for Traffic Simulation can be found on our Rating Policies page.
Overview of Rated Packages
We tested the following packages (click on name to get to the detailed rating information):
SUMO is definitely in the top league when it comes to open source implementations of microscopic traffic simulation. Its free availability, quick installation and excellent documentation make it easy for everyone to perform traffic simulation at city scale.
SUMO does not include nanoscopic modeling of participants, therefore the variety of actions that can be performed by providing discrete inputs (either vehicle controls or "commands") is limited.
The road / rail network and its attribution is key to what can be achieved by adding traffic participant distributions etc.
Even if you just want to add more traffic to your nanoscopic model, SUMO may be the way to go since it provides excellent means for extracting and injecting entities in full sync with its own simulation.
As a standalone tool, SUMO might be restricted to use cases in the domain of "planning and control". But interfaced properly with nanoscopic traffic solutions, it will add much value to other use cases like "perception", "decision making" etc.
This is definitely one of the most user-friendly tools which we have tested to date (July 2021). Support options are shown right on the landing page after you start the software. The user manual (1300+ pages) covers tons of features, use cases, and background information.
The road editor is easy to use and allows for the modelling of complex road networks based on map data (eg, by Bing, OpenStreetMap etc.), 3rd party networks (in OpenDRIVE format), or just from scratch.
The tool covers various mobility solutions; road vehicles as well as railroads and trams can be modeled including the respective signaling. Pedestrians can be added, including also features like boarding busses, trams etc. (for this, the Viswalk extension module is required).
PTV Vissim is a microscopic traffic simulation tool, therefore its strength clearly lies in the modeling of medium- to large-scale road networks with traffic flows, complex signaling etc. The modeling of individual vehicles with dedicated actions is not a built-in feature of this toolset (although users may address individual entities using the programming interfaces) but it would be a great extension and would relieve users from the need to couple Vissim with nanoscopic 3rd party simulation solutions. Adding a more sophisticated vehicle dynamics implementation would add quite some value, too.
The tool comes with good extensibility, plug-in features, and well-documented APIs (with lots of examples). Therefore, users may include Vissim into their own environments with moderate efforts.
In the ADAS/AD world, Vissim in its current state will, most probably, not be used standalone but always coupled with another simulation engine that provides vehicle dynamics, sensor simulation and nanoscopic traffic simulation capabilities. The harmonization of road networks for these coupled simulations is a key challenge, though. A full support of OpenDRIVE’s most recent version and an improved interpretation of OpenDRIVE junctions would make the coupling considerably easier.
Overall, the tool may well be on the doorstep for growing from a “nice-to-have” extension in ADAS/AD simulation to a key role in the creation and simulation of road networks and associated dynamic entities.
The current tool-suite seems to be a good fit for the following use cases:
algorithm development for ADAS/AD controllers
simulation for planning and control (object lists)
Standalone, it has limited value for the following use cases (which may be overcome by a coupling Vissim to dedicated 3rd party solutions):
creation of training data sets for perception systems
realistic sensor raw data generation and injection across all wavelengths
driver- and hardware-in-the-loop testing under real-time conditions
Please let us know if you are interested in having us review a specific product. Provided it fits our scope and in case it’s a free and open source product, things will be easy and we will be happy to execute your request asap. For products outside our scope we might need to discuss the relevance with you first. For fully commercial products we might need to contact the vendor and clarify access to further information, an evaluation copy etc. But we’ll happy to do this – it’s our task and “mission”.