Autonomous vehicle simulation is becoming an important part of how engineering teams develop and validate automated driving functions. It lets teams examine vehicle behavior in virtual roads and traffic scenarios, connect simulated data with control systems, and study interactions that are difficult to assess through physical testing alone. The technical program of the 2019 Driving Simulation & Virtual Reality Conference in Strasbourg highlighted several practical challenges for organizations building these workflows.
For automation engineers, system integrators and technical leaders, the central lesson is that simulation is not a single tool or test stage. Its value depends on how well models, scenarios, interfaces, standards and human factors are brought together across the development process.
Why autonomous vehicle simulation matters
Automated driving functions must be evaluated across a broad range of road conditions, traffic situations and system responses. The conference program described simulation as a way to conduct large-scale virtual testing, including for safety assurance. It also raised the question of how simulation can support the very large amount of driving needed to assess reliability, without suggesting that virtual testing removes the need for other forms of validation.
A useful simulation environment can bring together several elements: road networks, traffic participants, vehicle dynamics, sensors and the system under test. For example, a development team may combine a digital road model with a defined driving scenario and simulated sensor inputs, then observe how an automated driving controller responds. This helps make the test conditions explicit and repeatable.
Simulation can also support different points in the engineering workflow. The program included topics ranging from early algorithm testing and ADAS development to validation of highly automated driving functions, driver-in-the-loop studies and hardware-in-the-loop approaches.
Build a connected simulation workflow
Virtual testing is most useful when tools can exchange the information needed to represent the test and interpret its results. The conference’s standards discussions focused on simulation data interfaces, high-performance computing and the role of common formats in development and validation.
Examples discussed in the program included ASAM OpenDRIVE for road-network descriptions, OpenSCENARIO for dynamic driving scenarios and OpenCRG for road-surface representation. Open Simulation Interface (OSI) was also named among the relevant standards. These formats address different parts of a simulation problem; they are not interchangeable, but can contribute to a more consistent data flow between tools.
For engineering teams, the practical considerations include:
- Define what each model represents. Clarify the source and intended use of road, vehicle, traffic and sensor data.
- Check interfaces early. Identify how simulation tools, controllers and test systems will exchange inputs and outputs.
- Make scenarios reusable. Document scenario conditions so tests can be repeated and compared across systems.
- Plan for computing demands. Large scenario sets and detailed models can place demands on compute resources, making processing architecture part of the system design.
Standards can help improve integration and data exchange, but teams still need to verify that the selected formats and interfaces meet the requirements of their specific workflow.
Match test fidelity to the engineering question
Not every test requires the same level of physical detail. A simulation may provide ground-truth sensor information for algorithm development, or generate more realistic inputs—such as video streams for camera-based ADAS functions—for testing software that interacts with sensor data. The conference program also identified sensor models, raw simulated sensor data and hardware-in-the-loop testing as areas of interest for autonomous driving development.
The choice of fidelity should follow the question being asked. A team investigating a vehicle-control response may prioritize vehicle dynamics and road characteristics. A team assessing a perception function may need sensor outputs that are suitable for its software and test setup. When connecting simulation to real controllers or other hardware, interface timing and data handling become important elements of the test design.
Simulation platforms can also connect visual displays, motion systems, audio, control loading and driver inputs. These components contribute to the driver’s experience in a driver-in-the-loop study. Their importance varies by application: a study of vehicle handling, for instance, may require attention to motion cues and steering feedback, while an HMI evaluation may emphasize visual presentation and user interaction.
Account for human factors and the display environment
Human behavior remains relevant even as vehicle functions become more automated. Conference topics included driver conditions and transitions between automated and manual control, as well as perception, steering-wheel haptics, sound, vibration and motion cueing. These subjects show why a technical assessment of an automated function may need to consider both system response and the person interacting with it.
Virtual reality hardware also needs to be selected for its application. The program contrasted head-mounted displays with CAVE immersive rooms, noting that the systems differ and can complement one another. Rather than treating one display type as universally preferable, teams should consider the intended task, participant interaction and experimental context.
For driver-in-the-loop studies, consistent procedures matter too. The conference raised the need for guidelines that make results more comparable and transferable between simulators, alongside clearer approaches to classifying simulator systems and experimental practices.
Make virtual testing part of validation planning
Simulation can strengthen an engineering process when it is planned as one element of a broader development and validation strategy. That means selecting scenarios aligned with the function under test, documenting model assumptions, choosing appropriate sensor and vehicle representations, and checking how results move between tools and test environments.
For plant and engineering leaders overseeing automation programs, the practical implication is to evaluate the whole workflow—not just the simulator. A robust approach depends on compatible data, defined test procedures and an understanding of what each simulation result can and cannot establish. By combining those foundations with attention to human factors and hardware integration, teams can use autonomous vehicle simulation more effectively to develop and assess ADAS and automated driving systems.