In MEDITATE, we aim at improving co-simulation-based validation of ADAS with an integrated framework, supporting both the automated generation of a suite of test scenarios, and its optimization.

The main objectives of the project are:

• O1. Modeling Languages and Techniques for ADAS Testing Scenarios. We aim at identifying/defining a modeling language that allows for the representation of ADAS testing scenarios at different levels of abstraction, independently from their purpose or underlying technology.
• O2. Automated Generation of Testing Scenarios for ADAS. By leveraging the modeling languages and techniques defined in O1, we devise novel test generation strategies leveraging search-based approaches and/or AI techniques, such as Deep Reinforcement Learning.
• O3. Optimization of Testing Scenarios for ADAS. We aim at defining techniques to optimize the execution of test scenarios, by reducing the size of the test suite generated by O2 and prioritizing the most “relevant” test cases.
• O4. Integration of the developed solutions. A final objective is to integrate all the above described components within a coherent validation framework, empowering automotive car makers with a common infrastructure to support co-simulation-based validation of ADAS.