Autoware 2.0 moves the open autonomous-driving stack toward an AI-based architecture

What is changing?

TIER IV says it will lead Autoware 2.0 initiatives with the Autoware Foundation and the global open-source community. The architecture introduces a scalable framework for data-centric, AI-based approaches while preserving compatibility with conventional rule-based driving modules. The plan also includes an indexing scheme for combining modular software components and a Reference Design framework that can be adapted across vehicles and operating environments.

Autoware’s documentation stresses that version 2.0.0 contains only the first step of the roadmap: developers can select the learned Diffusion Planner at launch, but the rule-based stack remains the default. Later stages involving learned perception and end-to-end driving are still plans rather than current capabilities.

TIER IV aims to work with Arm on in-vehicle computing, AWS on cloud-native development, and Red Hat on AutoSD integration. The ecosystem could make it easier to commercialize open driving software across different hardware and operating conditions.