Autoware 2.0 lets open-source developers test learned driving models without abandoning the predictability of rule-based planning
The Autoware Foundation has released Autoware 2.0, extending its open-source autonomous driving software to end-to-end AI and hybrid stacks alongside its established rule-based modules. A package registry and reference design framework also arrive, helping developers turn components into complete vehicle systems.
At its core is a generator-selector setup: several trajectory generators run in parallel while a selector picks one. Developers can mix rule-based, AI and hybrid modules, testing learned planning against transparent rule-based code.
AI models now default to version-controlled storage on the foundation’s Hugging Face page. Autoware Index lists compatible repositories, and its command-line tool builds a workspace from chosen packages or a full reference design.
In a statement, Ryohsuke Mitsudome, Technical Steering Committee Chair at the foundation, said: “Users often ask us how to go from Autoware’s software to a complete autonomous vehicle. Autoware 2.0 addresses this by introducing Autoware Reference Design with a documented blueprint for the entire system. Through the Autoware Index, community members can share their packages, and users can now set up a reference design with a single command.”
Each design pairs a deployment profile for vehicle, hardware and use case with a maintained software stack, and must separate intended from proven capability. Designs span low-speed shuttles to urban robotaxis, and commercial partners can build their own stacks on top.
Source: The Autoware Foundation
Source: www.automotiveworld.com
