Autoware working group meeting 20190807

The next meeting of the Autoware working group will be held at Wednesday, August 7, 2019 2:00 PM. The meeting information is below.

On the agenda we have:

  1. Discussion of the specification-making process.
  2. Tasks for the first Autoware milestone

Please join my meeting from your computer, tablet or smartphone.
https://global.gotomeeting.com/join/604833597

You can also dial in using your phone.
United States: +1 (571) 317-3129

Access Code: 604-833-597

More phone numbers
Australia: +61 2 9087 3604
Austria: +43 7 2081 5427
Belgium: +32 28 93 7018
Canada: +1 (647) 497-9391
Denmark: +45 43 31 47 82
Finland: +358 942 72 1060
France: +33 170 950 594
Germany: +49 692 5736 7317
Ireland: +353 15 360 728
Italy: +39 0 230 57 81 42
Netherlands: +31 202 251 017
New Zealand: +64 9 280 6302
Norway: +47 21 93 37 51
Spain: +34 932 75 2004
Sweden: +46 853 527 827
Switzerland: +41 225 4599 78
United Kingdom: +44 330 221 0088

New to GoToMeeting? Get the app now and be ready when your first meeting starts:
https://global.gotomeeting.com/install/604833597

Reference standardisation processes:

All three are relatively light-weight processes. The IETF is the most tried-and-true for producing actual specifications of protocols and interfaces. The PEP process is also well-tested and battle-hardened, and the REP process effectively adopts the PEP process.

All three are based on the following basic concepts:

  • Anyone can propose something
  • Working code wins
  • Lightweight voting process for acceptance

In fact the processes spend more effort ensuring documents follow a good format, etc., than on prescribing a process.

The IETF process is the most well-defined, and includes detail on managing withdrawals, supersedings, conflicts, etc. It is probably too heavy-weight for us at this stage.

My recommendation is to adopt the PEP process with as few modifications as possible.

Sorry, I will be about 10minutes late to the meeting. I might be listening but wouldn’t be able to talk.

Attendees

  • Geoffrey Biggs (Tier IV)
  • Servando German (Linaro)
  • Ryohsuke Mitsudome (Tier IV)
  • Akihito Ohsato (Tier IV)
  • Maxim Osipov
  • Pedro Periera
  • Ahmed Radwan
  • Akihiko Tsukuda (eSOL)
  • Liyou Zhou (Arm)

Minutes Geoffrey Biggs

Discussion

Autoware specification process

  • eSOL: What changes would be required if we use the PEP process?
    • No major changes, but the PEP process is designed for a language, so we may need to make changes to what is contained in a specification document to account for this.
  • Arm: Why not start from the REP process?
    • The PEP process is more up-to-date and better structured.
  • If we want structure, why not use the IETF process?
    • It is too heavy for us at this early stage. We prefer to start with a light process and add to it as necessary rather than risk a heavy process choking off contributions.
  • Specification process definition work will be done in a repository dedicated to the specification.
  • The goal is to have a draft in place by September for discussion in this WG, followed by proposal to the TSC.

First Autoware milestone tasks

  • Tasks to be done for the first of the five milestones that are heading towards autonomous valet parking.
  • Linaro: Can we take existing source code from Autoware.AI to satisfy software tasks?
    • We can take algorithms, but the implementations need to be re-done to have tests, design documentation, user and developer manuals, be real-time safe (to the extent permitted by ROS 2), etc. Each algorithm needs to be implemented from scratch to meet these needs.
  • How are we going to deal with different vehicles?
    • We need to select a common vehicle platform. There is a task for this.
  • Not all companies have a vehicle.
    • Those without a vehicle will most likely be working in simulation, so they are good candidates to contribute to the simulation epic’s tasks.
  • AutonomousStuff already have a car that works with Autoware.AI. Do we need to redo their software?
    • Yes, it needs to be ROS 2-based, real-time, have tests and design documentation, etc. Hopefully AutonomousStuff will be willing to do this work. If not, someone else who is familiar with their hardware will need to do it.
  • The choosing the ECU task needs to coordinate closely with the reference platform WG.
  • Do these activities need to follow the specification process?
    • Yes
  • Does work that crosses over with other WGs need to follow those groups’ work or is it indepednent?
    • The other WGs are expected to work to these goals and follow the specification process. This is all one big project.

We are still in the early stages of this working group and haven’t settled on a regular meeting time yet; I hope we can do that in the next meeting. So to reach that point, we need to actually have that next meeting. Please fill in the following poll about your preferred time.

Please answer by next Friday (8/16) UTC.