Title

AWS re:Invent 2023 - How BMW and Qualcomm built an automated driving platform on AWS (AUT202)

Summary

• BMW and Qualcomm have partnered to develop an automated driving platform on AWS.
• Jörg Krebs from BMW introduced the session, followed by Sriram from Qualcomm and Maria from AWS.
• The platform is designed to handle the validation and verification of advanced driver-assistance systems (ADAS) by processing huge amounts of data.
• BMW's Neue Klasse, a new generation of cars starting in 2025, will feature electrification, digitization, and circularity, with a new IT and software architecture.
• Qualcomm contributes the Snapdragon Ride system on a chip and expertise in computer vision software.
• The platform is cloud-based to accommodate dynamic development processes, scalability, global collaboration, and reduced time to market.
• AWS was chosen for its managed services, technical expertise, and ability to quickly set up the platform using existing architectures and blueprints.
• The platform's design principles include reusability, collaboration, seamless workflow for various personas, and global scalability.
• The architecture is layered, with AWS services at the base, followed by open-source capabilities, common capabilities, and independent realms.
• The platform features data ingestion, cataloging, enrichment, labeling, simulation, verification, validation, and advanced analytics.
• Qualcomm's AI100 accelerators are available on AWS for running advanced AI models.
• Maria detailed the development journey on the platform, focusing on data, infrastructure, development acceleration, realm communication, and data visualization.
• Key takeaways include the platform's self-service UI, infrastructure as code, offloading of undifferentiated heavy lifting, acceleration of development, and flexible, standardized communication between realms.

Insights

• The collaboration between BMW and Qualcomm on AWS showcases the automotive industry's shift towards cloud-based platforms for developing advanced technologies like ADAS.
• The platform's emphasis on reusability and collaboration indicates a trend towards more modular and interoperable systems in software development.
• The use of AWS managed services and infrastructure as code suggests a growing reliance on cloud providers for not just storage and compute, but also for the rapid deployment and scaling of complex systems.
• The integration of Qualcomm's AI100 accelerators into the AWS platform highlights the increasing importance of specialized hardware for AI and machine learning workloads in the cloud.
• The session demonstrates the complexity of developing ADAS features and the need for sophisticated tooling and infrastructure to manage the lifecycle of such applications, from data ingestion to simulation and validation.
• The global scalability of the platform reflects the automotive industry's need to develop and test ADAS features across diverse driving conditions and regulatory environments worldwide.
• The platform's ability to offload undifferentiated heavy lifting and accelerate development aligns with the industry's goal to reduce time to market for new vehicle features and technologies.