Engineering best practices for algorithms virtual validation of autonomous vehicles

One of the main challenges in ADAS and autonomous vehicle development is the validation of the perception, planning, and control methods and algorithms. Environment and driving scene recognition are carried out using the data from several sensors and camera images, fused together in a proper way. Deep learning networks are trained, based on data from test drives. A wide variety of traffic scenarios must be considered in order to ensure sufficient coverage. These scenarios are also used for the validation of the safety of control algorithms.

Virtual validation and testing are used to achieve these requirements in a reasonable time frame. A virtual framework is created, including representation of the sensors and traffic environment, together with realistic vehicle dynamics as often required.

In this webinar, our experts introduce and illustrate the different steps from the “sensing” of the environment up to the definition and tracking of a suitable vehicle trajectory. It covers:

• Virtual scenarios to expand coverage and improve safety verification
• Main requirements of the virtual testing framework
• How to achieve a good balance between all attributes including comfort, performance, fuel economy, etc.