Open AD Kit WG Meeting 2022/04/14

[ghost]

Administrative

Previous meeting minutes

Attendees

Chaired by David

• Armağan Arslan (LeoDrive.ai)
• Christophe Le Dorze (SUSE)
• Daisuke Hashimoto (T4)
• David W (A3VG)
• Eiji Sekiya (T4)
• Fumihito Ito (T4)
• Hiroshi Igata (T4)
• Kei Kato (T4)
• Lalith Vipulananthan (T4)
• Naoyori Tanzawa (T4)
• Ryohsuke Mitsudome (T4)
• Yoshiya Fujii (T4)

Agenda

• Introduction of new members
• Announcements
• Review of backlog actions
• Review of open discussions
  • KLT based Multi Tracking #116
  • https://github.com/autowarefoundation/autoware/discussions/89
  • Open AD Kit draft schedule
• Discussion topics
  • Open AD Kit GTM Proposal (CJ)
  • Volunteer call: Engineers with K3s/orchestration experience wanted!
• Q&A follow-up
  • Update how to get Autoware #127
  • bug fix #126

Introduction of new members

• Desh (Marvell) - has attended previous WG meetings and is back after a break.

Announcements

Review of backlog actions

• Create a discussion for the Open AD Kit WG to discuss data recording requirements with the Software WG (eg: which topics are required for a minimum data set) @mitsudome-r - due 04/13
  • Done! Develop multi lane #166
• Ask ITRI if they are willing to provide map data for their private campus @Igata-ctrl - due 04/15
  • Done! Private campus map data has been provided and is available for use with T4's CI/CD system
  • Currently in discussion to see if ITRI will share the map data without restrictions
• Ask Excelfore to clarify the license that will be provided with the eSync Autoware PoC @LalithVipulananthan - due 04/08
  • Done! License question was added to the existing Q&A discussion and currently pending a response

Review of open discussions

Relationships between artifacts in AWF working groups

• Aiming to close out the discussion - no comments from T4 nor any other attendees.

Open AD Kit 2.0 scope

Schedule (Gantt diagram)
https://docs.google.com/spreadsheets/d/1z56AOAaZB2wUkwZv5GZy6BrN9bUgbX6upFLEW7-5sz8/edit#gid=1115838130

• David summarised the CI/CD sync meeting earlier in the week, where there was an effort to clarify the purpose and meaning of Open AD Kit (since there is some confusion about the original Open AD Kit vision in terms of infrastructure)
  • Hashimoto-san's OpsCon/ConOps proposal for defining Open AD Kit 2.0 scope is the application/methodology side of things
  • Kasper's work is on the infrastructure (or "plumbing" side)
  • WG members are looking for something demonstrable with the ITRI bus project, but need to also focus on Open AD Kit infrastructure (eg: containerization)
  • Fatih is currently working to list all ROS nodes (inputs and outputs), then grouping them into clusters with the correct runtime for each group. This list plus the architecture diagram Fatih created could be used to determine which nodes go into which containers as a possible reference.
• Tanzawa-san commented that use cases and benefits still need to be determined, and the data recording work should be be done in parallel
• Armağan noted that data recording is necessary, with or without Open AD Kit, and suggested Mapping as a possible component for a containerization demo, as it is a separate function and so could be a better demonstration of containerization (eg: TomTom's map loading component, demonstrated in a previous meeting). With current map functionality in Autoware, we need to be able to delete map data behind the vehicle and load map data in front of the vehicle. Not quite Open AD Kit, but could demonstrate updating map data for all vehicles.
  • Mitsudome-san said that TomTom does not provide pointcloud maps at the moment through their map service. Working on this as a PoC but still in early stages and was not sure if map data is something that should be containerized (Armağan noted that it was just an idea) but thinks that we need to consider how we demonstrate Open AD Kit
• CJ suggested localization for a demo of containerization
  • 3D lidar for bus ODD, running on the bus ODD hardware spec, and 2D Lidar for F1tenth, running on NVIDIA Jetson
  • Demonstrates ability to tailor different containers for your purpose, demonstrate portability across different hardware
  • Feedback from F1tenth community is that Autoware is too complicated to run on their hardware, so we need to provide something that is more useable
  • Mitsudome-san noted that the sensor setup is different for each case, so portability is not necessarily being shown, but something to consider
• Next steps:
  • Fatih to create a complete list of all topics/nodes in Autoware
  • Discuss candidates for containerization

Open ADKit Slide Reviews.pdf

• Sekiya-san presented his feedback on Open AD Kit 1.0 and scope ideas for Open AD Kit 2.0
  • Autoware should be available in both monolithic and containerized forms
  • For software developers, they can still continue with Autoware even as a monolith, so we need to explain the benefits to them to get them onboard with microservices and orchestration
  • CI/CD pipeline set up will require a lot of resources, and we need to understand what will be needed to achieve this goal
• Kasper agreed that it should be possible to run Autoware without an orchestrator, and that the planned infrastructure will allow for Autoware to be built and run as a monolith. However orchestration knowledge is lacking within AWF.
• David reached out to SUSE to get their thoughts on helping the WG with orchestration related tasks, but they were not able to join. Also need to get buy-in from the software group

Discussion topics

Open AD Kit GTM Proposal

• Postponed for now at CJ's request

Q&A follow-up

SUSE

• No additional comments or questions.

eSync

• No additional comments or questions.

Action Items

• [ ]

[ghost]

@xmfcx @mitsudome-r Just to check, is there an issue tracking the task to create a complete list of all topics/nodes in Autoware?