Enabling In-Vehicle Intelligence: Closing the Edge AI Loop with Unified Software Lifecycle and Data Management
Title: Enabling In-Vehicle Intelligence: Closing the Edge AI Loop with Unified Software Lifecycle and Data Management
Short Description:
Description:
Today’s software-defined vehicles depend on reliable data flows and agile software systems. But in practice, many embedded AI deployments struggle due to fragmented infrastructures: data management and software updates are handled in silos, limiting intelligence, increasing latency, and weakening system robustness.
This session presents a modular and field-tested approach to unify these processes by closing the Edge AI loop in the vehicle: software deployment (applications, edge AI agents, or ML models); collection and selection of valuable vehicle data at the edge; continuous learning; model and agent optimization, redeploying to improve the system and edge architecture.
With each cycle, the in-vehicle intelligence becomes smarter, enabling moving from isolated pilots to robust real-world AI solutions.
Attendees will gain insights into
• How to run the full AI lifecycle, turning pilots into scalable, production-ready solutions;
• How to deploy, update, and monitor applications and ML models across distributed invehicle systems;
• Selective, privacy-respecting data collection and routing in resource-constrained environments;
• How to use open standards and collaborate with partners to cut costs for software development.
• How to ensure memory safety in SDV applications to prevent security vulnerabilities and protect data, models, and software across the lifecycle.
• Lessons learned from real-world implementations in regulated industries like automotive.
Consolidating data flow and software control creates adaptive systems that improve over time—securely and at scale. This talk is non-promotional, hands-on, and designed to inspire collaboration on new architectures for data-driven mobility.
Takeaways
1. Understand the concept of "closing the AI loop" in embedded automotive systems.
2. Learn from real deployment experiences and security challenges in vehicle-edge environments.
3. Explore architectural strategies for unifying data pipelines and software lifecycle management