Autoware Labs Proposal

[xmfcx]

We are excited to share with you the Autoware Labs proposal, aimed at addressing key organizational and operational challenges within the Autoware Foundation.
Please review the comprehensive overview provided in the slides at this link.

After viewing the slides, we invite all members of the Autoware community and AWF member companies to provide their feedback and insights on this proposal. Your opinions are essential for refining our approach and ensuring the success of Autoware Labs.

How to Provide Feedback

• Comment on this Discussion: Share your thoughts directly in the comments section of this GitHub discussions post.
• Join the Discussion at the TSC Meetings: This proposal will also be a topic of discussion in the upcoming Technical Steering Committee (TSC) meetings. We encourage your participation.

Quick Outline from the Slides

Organizational Problems

• Lack of a coherent vision
• Very hard to make changes in Autoware (and AWSIM)
• WG leaders can’t assign tasks

Ideas to Mitigate Contribution Friction

• Identify and resolve the source of blockers
• Maintainer of the blocking side should open up for quick conversations
• Track blocked issues/PRs in an Epic issue and discuss them in the monthly TSCs
• Move or Fork dependent repositories to the Autoware Foundation GitHub organization

Request

• Formation of a focused Autoware Labs team
• Set the goal to achieve Dense Urban ODD
• Track blocked issues/PRs and raise them to TSC
• Move or Fork blocking repos to AWF org

Key Actions

• Development of evaluation criteria
• Creation of Autoware Labs
• Coordination within existing Software Working Group
• Resource allocation

We are committed to transparency and collaboration in this initiative. Your feedback will play a crucial role in shaping the future of Autoware Labs and ensuring it meets the community's needs and expectations.

[idorobotics]

Please also complete the Autoware User Survey.

[doganulus]

Some problems mentioned in the presentation can be avoided with better organization. If the newly formed group, or another dedicated group, takes over the interface management and iteration across the Autoware ecosystem, that would help reduce the friction considerably. Especially if the working groups are assigned to target interfaces within the given timeframes.

Some other challenges, like effective testing and evaluation, require many breakthroughs on multiple fronts. The efforts must range from ROS's inner workings (https://github.com/orgs/autowarefoundation/discussions/2673) to containerization efforts (tier4/AWSIM#7) and the establishment of safety culture across the organization. Since you have mentioned, the evaluation criteria must be changeable and iterable, and focusing on the simulation infrastructure that can handle arbitrary metrics extracted from simulations is better than committing to concrete metrics. ASAM's report on the topic can be taken as a reference.