Autonomous Driving Validation Library – A New Way to Test and Validate Autonomous Vehicles

As the world moves towards a future of autonomous vehicles (AVs), it becomes increasingly important to ensure that these vehicles are safe and efficient. To achieve this, commercial companies and researchers worldwide are developing new technologies and methods to test and validate AVs. One such solution is Cognata’s Autonomous Driving Validation Library (ADVL), a combination of a scenario-based simulation engine and a comprehensive library of scenarios aimed at testing and validating AVs through standard packages of different driving scenarios.


ADVL building blocks logic starts from basic scenarios that test, for example, lane keeping during a bend, Automatic Lane Keeping System (ALKS) or Electronic Stability Control (ESC), or avoiding different types of safety hazards on the road, such as Emergency Brake Assist (EBA). These scenarios then gradually progress to test the AV in more challenging and complex situations and events, such as Lane Support Systems (LSS) or Adaptive Cruise Control (ACC). The ADVL library is based on growing worldwide research that aims to develop safe and effective autonomous driving, promote AV validation, and enable AV regulation.


Other companies are also investing in the development of AV technology and testing methods. Some have built testing facilities, where they test AVs in a range of scenarios, including simulated city environments and challenging weather conditions. Some rely on real-world data collected from their customers’ vehicles to continuously improve their Autopilot system. Companies and consortiums alike have been testing AVs in multiple cities across the world, progressing the autonomous vehicle ambition.


In addition, some countries have been actively promoting AVs and have established regulations; Korean Recommendations for Autonomous Driving System (K-RADS), the National Highway Traffic Safety Administration (NHTSA) guidelines in the United States, the Euro NCAP ratings in Europe, and the Japan ADAS regulations, all serve as existing and inspirational benchmarks for AV testing and validation.

Paris 1 copy 1024x576 - Autonomous Driving Validation Library - A New Way to Test and Validate Autonomous Vehicles


Despite the growing interest in AV technology, there are many challenges that companies face in testing and validating AVs. One of the biggest challenges is creating realistic and diverse scenarios for testing. AVs need to be tested in various scenarios, from simple driving tasks like lane keeping to more complex situations like navigating through heavy traffic or extreme weather conditions. Creating these scenarios requires a deep understanding of human driving behavior and the ability to simulate a wide range of driving conditions accurately.


Another challenge is ensuring data accuracy and consistency. AVs generate large amounts of data, including sensor data, GPS data, and mapping data, which need to be analyzed and validated to ensure its accuracy and consistency. This requires sophisticated algorithms and machine learning techniques to identify and correct errors in the data.

Finally, managing the large amounts of data generated by AV testing is a significant challenge. AV testing generates terabytes of data every day, which needs to be stored, processed, and analyzed in a timely and efficient manner. This requires high-performance computing infrastructure and sophisticated data management tools to ensure that the data is properly organized and accessible when needed.


At Cognata, we understand the importance of automating and streamlining the time-consuming and computing-intensive validation process. That’s why we created easy-to-use scenario packages that are simple to run through our platform with the client AI or to generate datasets. We’ve also added automation in the form of dynamic analysis rules (DAR) that provide automatic analysis based on configurable sets of logical parameters.

The ADVL scenarios were created in different photo-realistic highway and urban scenes and included various models, dynamic road users, static objects, and variables like different speeds and locations. These scenarios can be tested in a range of conditions, including harsh weather conditions, such as heavy rain, visual challenges, and degraded road conditions, such as potholes, gravel, or dirt roads. The ADVL scenarios can also be tested at different times of day, including low-light or nighttime conditions, which leads to hundreds and even thousands of simulations for each ADVL package. This allows for large-scale simulation that will take your technology to the next level of autonomous driving.


Using ADVL, you can ensure that your AV technology is safe, reliable, and effective. Our solution helps you test and validate your AV in a range of scenarios, from simple to challenging, and in a variety of different conditions so that you can be confident in its performance on the road. By addressing these challenges and leveraging advanced technologies like machine learning and scenario-based simulation, companies can ensure that their AV technology is ready to hit the road confidently and help usher in a new era of autonomous mobility.