Behind the scenes | Prospective Safety Assessments

Within the V4SAFETY project, we provide a framework for prospective safety assessments using virtual simulations.
To ensure stakeholders can confidently rely on the outcomes of these simulations, validation and verification (V&V) of results are essential. V4SAFETY therefore integrates a dedicated V&V process tailored to the various stages of a safety assessment, with the goal of enhancing trust and credibility in simulation outcomes.

Validation and verification framework
The V&V process in V4SAFETY defines a systematic procedure for evaluating the validity of results at each stage of a safety assessment. It incorporates both qualitative and quantitative methods, explicitly linked to the corresponding assessment phases. To support practical implementation and provide clear guidance, V4SAFETY has also developed a set of question cards that highlight potential pitfalls and challenges, helping practitioners avoid common errors. These question cards will be published in Deliverable D2.1.

A key validation method involves comparing simulation results with experimental data gathered in controlled environments, such as test tracks.

Real-world validation at the Aldenhoven Testing Center
To demonstrate this validation approach, BASt and TNO jointly conducted tests at the Aldenhoven Testing Center on 28 and 29 April 2025. The tests aimed to compare simulation outputs with real-world data across two essential use cases:

Use case 1: brake model validation
This test validates a braking model within a pedestrian scenario. In simulations, the Autonomous Emergency Braking (AEB) system triggers a full braking manoeuvre to avoid collisions. The primary goal is to compare key parameters - such as maximum deceleration, deceleration ramp-up, and braking distances - between simulations and controlled braking tests on the test track.

Use case 2: sensor model validation
This test evaluates the detection capabilities of sensor systems in scenarios involving a Vulnerable Road User (VRU) crossing the street. In simulations, an ideal sensor model is used to detect pedestrians, whereas real-world sensor systems have inherent limitations. The aim is to compare simulation results with data from a production-series sensor system to assess discrepancies between idealised and actual conditions. Key parameters include pedestrian detection timings and the influence of varying vehicle speeds on detection performance.

For these tests, the TNO CarLab - a modified production-series vehicle equipped with advanced sensors, computing, and communication systems - was used. BASt provided complementary vehicle safety test equipment, including a dummy propulsion system.

Future outlook and reporting
Data from these validation tests are currently being analysed. Comprehensive findings and insights will be presented in Deliverable D2.2, part of Work Package 2; Prospective Safety Assessment Framework.

The video shows a top view of a test run in which the AEB system of the vehicle under test reacts to a pedestrian crossing the road.

The video shows a front-view of a test run in which the AEB system of the vehicle under test reacts to a pedestrian crossing the road.