ADAS Validation

From driver assistance systems to automated driving: technical development is advancing at a rapid pace and we are seeing a continuous increase in both the degree of connectivity and the number of functions being implemented. In turn, the validation of driving functions is becoming equally more complex. Bertrandt is specialized in the validation of physical driving functions that relieve the driver – even as far as allowing the vehicle to take over driving completely.

The focus of our validation team is on three main areas: comfort functions such as Adaptive Cruise Control (ACC), active safety functions Warn, Brake, Avoid (WBA) and chassis control systems (Electronic Stability Control – ESC). Furthermore, we are also responsible for validating the actuators of the braking system.

Our team provides robust validation of driving functions on the component, system, and vehicle levels. These tests are carried out either entirely virtually or in the vehicle, or as a combination of both. We already support the creation of the necessary software by carrying out simulation-aided tests – as soon as the systems and requirements have been clarified. We take responsibility for all process steps, from the implementation and concept development of test cases to validation on the test bench and ultimately to a recommendation for approval. Our aim is always to fully comply with all legal requirements and to safely achieve consumer protection targets, e.g. EuroNCAP.
At our testing facilities, we validate fundamental sensor functions as well as their final calibration in the complete vehicle under different weather conditions. System tests for the comfort functions in particular are already performed at a very early stage on the test track or on public roads. On the test track, we can flexibly represent and implement the test scenarios with a high level of precision using our x-track mobile laboratory.

Whether it is Autonomous Emergency Braking (AEB), Emergency Lane Keeping (ELK), vehicle dynamics or deceleration management, the efficient validation of each driving function requires a profound understanding of functions and systems. We have this extensive expertise at our disposal in-house and can interconnect different areas of function and system understanding at any time – and always for the benefit of our customers. In particular, we cooperate with the areas of Software Experts, Autonomous Mobility and Information Systems, and Electronics and Virtual Testing Solutions. Such interdisciplinary interaction is routine practice in our expert organization.

The number of sensors in the vehicle is constantly increasing. As a result, we need to merge the data streams from many different sensors. We test the interaction between the laser scanner, radar, ultrasound, camera, and lidar systems at the interfaces. This lets us find out which of the sensors has, for example, not recognized a pedestrian or cyclist, thus enabling us to evaluate and analyze improvements in the performance of the complete system.

Learning systems based on artificial intelligence (AI) are the future. This is now a certainty, not only since the introduction of legislation for Level 4 autonomous driving. Rule-based systems are still dominant. But wherever neural networks or deep learning are involved, we are already dealing with learning systems. We have been systematically expanding our validation expertise in this field since 2018. The reason is that neural networks and deep learning must be evaluated and validated in a fundamentally different way. Bertrandt recognized this early on and has demonstrated its competence with its HARRI innovation platform for the “Park and Charge” use case. With our team, we played an essential role in the field of 2D/3D object recognition.