Transforming AV/ADAS development at Continental with generative AI (AUT307)
Autonomous Mobility Development with AWS and Continental
Key Takeaways
Challenges in Autonomous Mobility Development:
Unprecedented data volume (up to 100 PB per year)
Diverse data requirements across development teams
Scalability and cost challenges as development programs change
Integrating various tools and applications
Continental's Collaboration with AWS:
Developed a platform called "CATCH" (Continental Automotive Edge Computers) on AWS
Leveraged AWS managed services to store and process data
Built a set of data services to enable data-driven development
Integrated with various partners and tools to reduce integration time
Scene Search:
Developed a scene search feature to help developers and engineers quickly find relevant driving data
Leveraged state-of-the-art approaches like generative AI, machine learning, and semantic search
Provided a user-friendly interface for data curation and visualization
Challenges in Autonomous Mobility Development
Autonomous Mobility development involves a significant shift from level 2 to level 3 automation, where the vehicle takes more control.
Key challenges include:
Unprecedented data volume (up to 100 PB per year) from sensors like cameras, radars, and LiDARs.
Diverse data requirements across development teams (labeling, ML training, simulation, verification, etc.).
Scalability and cost challenges as development programs change (e.g., from robo-taxi to passenger vehicles).
Integrating various tools and applications required for different tasks.
Continental's Collaboration with AWS
Continental is a major tier-one supplier in the automotive industry, with experience in developing autonomous driving features for over 20 years.
To address their challenges, Continental collaborated with AWS to build a platform called "CATCH" (Continental Automotive Edge Computers), hosted on AWS.
CATCH provides the following capabilities:
Ingesting up to 1 PB of data per day and processing it.
Offering data services and workbenches for developers to access and use the data.
Enabling third-party connections for data labeling, simulation, and other tasks.
Key targets for the collaboration were:
Reducing operational costs by 60%.
Reducing the time developers spend on data curation and discovery.
Improving the ability to integrate new technologies and partner solutions.
Scene Search
One of the outcomes of the collaboration was the development of a scene search feature.
The scene search feature aims to help developers and engineers quickly find relevant driving data for their tasks, such as:
Semantic search using natural language queries (e.g., "find pedestrians in the rain").
Similarity-based search using images or video snippets.
Structured search combining various criteria (e.g., object type, weather conditions).
The scene search feature leverages state-of-the-art approaches like generative AI, machine learning, and semantic search to provide a comprehensive data discovery experience.
It also includes a user-friendly interface for data curation and visualization, allowing developers to quickly identify true positives and false positives in the search results.
Call to Action
Reach out to an AWS representative to learn more about the collaboration approach and how it can benefit your autonomous mobility development efforts.
Provide feedback on the session to help improve future presentations and demonstrations.
These cookies are used to collect information about how you interact with this website and allow us to remember you. We use this information to improve and customize your browsing experience, as well as for analytics.
If you decline, your information won’t be tracked when you visit this website. A single cookie will be used in your browser to remember your preference.