Automotive Grade Linux All Member Meeting Spring 2025

In recent years, the core of innovation in automotive system architecture has been the integration of high-performance computing (HPC) as Vehicle Compute. This involves consolidating various software that was previously distributed across multiple ECUs onto the HPC. This is not only a hardware architecture innovation but also requires a significant overhaul of the existing ECU software architecture. For traditional automakers with a large legacy of automotive control software assets, this presents a major challenge. Many companies have started experimenting with various approaches to achieve Software-Defined Vehicles (SDV), but it is not always straightforward to progress towards their goals.

In this session, we aim to share ideas on the difficulties automakers face in adopting HPC and how AGL can potentially contribute to solving these challenges.

This session explores the integration of LLM (Large Language Models) and autonomous vehicle technology. It covers how vehicles can interpret and execute complex driving commands given in plain text, such as "Drive to the nearest gas station" or "Park in the designated spot." The session highlights advancements in machine learning and AI models that enable vehicles to understand and act on human instructions, pushing the boundaries of human-vehicle interaction.