Problem Statement

The development and validation of AV systems require comprehensive testing across a wide range of real-world scenarios to ensure robustness, safety, and reliability. Traditional scenario generation methods are often manual, time-consuming, and may not cover the vast spectrum of real-world variability and complexity. There is a a need for a advanced scenario generation approach that:

• Systematically integrates data from diverse sources, including high-quality datasets like NuScenes/NuImages, Cityscapes, BDD100K, Mapillary, and the EuroCity Persons Dataset.


• Leverages AI and Monte Carlo simulations to automate and enhance the generation of logical and concrete scenarios. These scenarios must be derived from functional scenarios aligned with our Operational Design Domain (ODD), encompassing diverse conditions such as urban environments, weather variations, pedestrian interactions, and rare or hazardous situations.

Objectives

• Develop and implement an innovative solution that automates the generation of detailed and realistic driving scenarios for AV testing, using AI and Monte Carlo simulations.


• Analyze and leverage large datasets from diverse sources to extract relevant parameters and conditions that impact AV operation.


• Ensure seamless integration of functional scenarios into the scenario generation process, maintaining alignment with our ODD.


• Collaborate with cross-functional teams to apply generated scenarios in testing protocols, facilitating the iterative improvement of our AV systems.

Expected results

• Analyze and preprocess datasets from leading sources like NuScenes/NuImages, Cityscapes, BDD100K, Mapillary,etc. to identify patterns and parameters for scenario generation.


• Contribute to the development of AI models and Monte Carlo simulations for expanding functional scenarios into logical and concrete scenarios, ensuring they encompass the required diversity and complexity.


• Utilize Monte Carlo simulations to explore variations within these scenarios, creating a comprehensive set of conditions for testing.


• Ensure seamless integration of functional scenarios into the scenario generation process, maintaining alignment with our ODD.


• Participate in the evaluation of test results, leveraging data to refine and improve the scenario generation process.

Join us in Internship!

CEA Tech Corporate from CEA Tech on Vimeo.

As an intern at CEA, you will have the opportunity to work in a world-renowned research environment. Our teams are made up of passionate and dedicated experts, providing a framework conducive to learning and collaboration. You will have access to cutting-edge equipment and top-notch research resources to carry out your missions.

• Student or recent graduate in computer science, engineering, data science or related field.


• Skills in AI, machine learning, statistical analysis and data processing (e.g. proficiency in Python, R, MATLAB).


• An interest in autonomous vehicles, robotics and the application of AI to solve complex real-world problems.


• Problem-solving skills, attention to detail and commitment to quality.


• Excellent communication skills and ability to work effectively in a collaborative team environment.