July 17, 2019, 22:00 UTC
- Lee Baldwin (AutonomousStuff)
- Alfredo Bencomo (Open Robotics)
- Geoffrey Biggs (Tier IV)
- Cheng Chen (AutoCore)
- Kenji Funaoka (Tier IV)
- Brian Holt (Parkopedia)
- Shinpei Kato (AWF)
- Seonman Kim (LGE)
- Dejan Pangercic (Apex.AI)
- Paul Sastrasinh (TRI-AD)
- Antonis Skadasis (StreetDrone)
- Stephane Strahm (Kalray)
- Akihiko Tsukuda (eSOL)
Minutes Geoffrey Biggs (Tier IV)
- Opening remarks and new member introductions (Board)
- Admin Confirmation of previous minutes (All)
- Follow-up Action items from previous meeting (All)
- Admin Names and photos of TSC members for the AWF website
- Admin Autoware annual summit
- Working group report Autoware (@gbiggs, @esteve)
- Working group report Map formats (@Brian_Holt)
- Working group report Vehicle interfaces (@Antonis_Skardasis, @Lee_Baldwin)
- Working group report ECU/Platform (@sstrahm)
- New working group proposals Simulation
- Planning Hiring a system architect for Autoware.Auto (Apex.AI/Tier IV)
- Release the safety case example
- Start a Discourse poll on when to hold the Autoware Summit
- Begin scoping the work and skills for an autonomous driving system architect
Opening remarks and new member introductions (Board)
- New observer: Jit Chowdhury from Auro/Ridecell
Confirmation of previous minutes (All)
- Minutes approved
Follow-up: Action items from previous meeting (All)
- Close remaining Slack channels except general
- Action taken Channels closed
- Follow up about free AWS minutes
- Geoff & Esteve & Jan
- Action taken Jan is setting up a meeting to talk to AWS
- Provide a set of launch files for starting just waypoint following-related functionality in Autoware.AI
- Action taken A repository has been created (please send link). This work is being done closely with StreetDrone.
- Start OpenDrive discussion in the Autoware Discourse category
- Action taken Not started; will be done as part of the maps working group.
- List planned contributions that go towards Foundation work and can be used to meet the milestone demos. Place planned contributions in this topic.
- Action taken Planned contributions listed
- Begin coordinating working groups
- Action taken Four working groups have been formed and starting working
- Safety case for autonomous driving in constrained evironments (i.e. autonomous valet parking) sample
- Action taken There are some legal problems with releasing the safety case documentation; the group who made the documents are worried about liability so Brian is working on persuading them.
- The Autoware working group held its first meeting on 16th of July
- The Autoware working group is proposing that its main output be a specification that defines what Autoware is
- Autoware.AI 1.12 was released.
- There are known bugs so a patch release is planned.
- The first release of Autoware.Auto (named Axle) has been made.
- Proposal The Autoware working group proposes that the Autoware Foundation produces a specification defining Autoware as one of its primary products, with Autoware.Auto being the reference implementation of this specification. All working groups will work towards producing a part of this specification, with the Autoware working group being responsible for cross-cutting aspects and the overall specification.
- AutonomousStuff: The specification idea is the right way to move forward, but what about Autoware.AI?
- It will not be forcefully aligned with the specification because it is due to be gradually phased out. We may need to alter Autoware.AI to match the specification in places to make the bridge work, but we want to keep these to a minimum because any resources put into Autoware.AI development are resources taken from Autoware.Auto development.
- Apex.AI: Will the specification have functional units or just high level stuff.
- We will be defining things at the ROS interfaces level and the architecture level, i.e. nodes.
- Apex.AI: Will this be a code-first or specification-first approach?
- A hybrid, iterative approach. We will design the specification to achieve some capability, then implement to prototype and check, then revise the specification.
- Proposal approved. Autoware working group will begin preparing a process for the specification work.
Working group report: Map formats (@Brian_Holt)
- First meeting was held just a few hours ago.
- There was initially quite a lot to discuss. (Minutes available …)
- Next meeting will be held in a week.
- Will meet every week for the next few weeks until the direction is set, then will probably drop down to every two weeks.
- Before the next meeting, the group will analyse a set of map formats and understand how they are useful, what they offer, etc.
- Apex.AI: Will there be different formats for simulation and on-vehicle software? Will the working group answer this question?
- The working group will definitely answer this question, especially because the current expectation is that simulation will use OpenDrive extensively, while on-vehicle will use a more suitable format.
- The fact that most people do not have access to the Navigation Data Standards Association specifications makes it hard to use the data widely. Converters will be available.
- Apex.AI: Is Lanelet2 still in the race?
- We need to separate out the storage format from the libraries/abstractions/APIs that we use to interact with map data; Lanelet2 is the latter. However Lanelet2 has become very closely coupled to the underlying storage format they currently use (OpenStreetMap), so work needs to be done to reduce this coupling.
- Apex.AI: Will the evaluation will be done at a theoretical level or will some practical testing be done?
- There are some sample maps available that can be used to assist with the evaluation, but there are not enough. We will need to refer to other evaluations as well as do our own work.
Working group report: Vehicle interfaces (@Antonis_Skardasis, @Lee_Baldwin)
- Second meeting was held last week.
- Attended by AS, StreetDrone, Apex.AI and Tier IV
- Use cases from each member have been collected.
- The next step is to start working on a common design to satisfy these use cases.
- Apex.AI: Comment on the use cases
Working group report: ECU/Platform (@sstrahm, @cheng.chen )
- Working group was launched with a call for participants, but first meeting has not yet been held.
- Working group outline is being prepared
[WG RP] Autoware Reference Platform Definition Documentation
- Tier IV: There will be more ECU vendors and silicon vendors joining the foundation, so we need to clarify what they can contribute to the foundation in this working group. Their roles in the foundation (not the market) need to be clear.
- This will be clarified in the first call.
- Apex.AI: Please think about hardware support for real-time/time determinism/other safety related stuff.
New working group proposals
- Tier IV proposes a new working group to deal with matters relating to simulation.
- @Masaya_KATAOKA proposed as one leader.
- Parkopedia: There will be some cross-over with the map WG, e.g. does the map provide enough information for the simulator? However in general having a WG to give direction to simulation will help with this.
- AutonomousStuff: Agree with having a simulation WG to help with the V&V aspects of Autoware. Metamoto is a company that could be interested in being involved.
- Tier IV: Having coordination of the various different simulators would help to understand the strengths and weaknesses of each.
- Tier IV: The WG could put some effort into producing scenarios (e.g. EURO NCAP using OpenScenario) that can be used widely in different simulators.
- Apex.AI/Parkopedia: We are working on carpark maps, so this could be something to provide simulation worlds for.
- Parkopedia: There seems to be a general movement towards Carla; the IV 2019 workshop had two presentations about Carla.
- LGE: We are interested in a simulation working group, and Dmitry can be the main contact for that, and possibly co-leader.
- Will start a simulation working group.
Hiring a system architect for Autoware.Auto (Apex.AI/Tier IV)
- Although we have a lot of engineers avialable and the two leaders of Autoware are skilled in software development for critical systems, we don’t have a system architect who is experienced in autonomous driving. We need someone who can do things like choose the best algorithms and understand what the interfaces need to do.
- For example, the current default path following algorithm is pure pursuit, but it is definitely not the best. MPC is being introduced but it is still not optimal.
- Therefore we should find someone who has this experience and knowledge and employ them full-time to do this work for Autoware.
- StreetDrone: It would be useful to have someone to do the top-level design, but we need to carefully specify what this person would need to do.
- Parkopedia: Would this person be employed directly by the Autoware Foundation, or be employed by a member company to work 100% for the AWF?
- Apex.AI: No opinion, but not sure if the AWF can even employ someone. So probably best would be to start through one company.
- Tier IV: We don’t even know if there is such a person available because autonomous driving technology is so closed. But we agree on the need for someone to be in charge of the architecture full time.
- AutonomousStuff: Agree completely because you need to have someone full-time on the system architecture who can make binding decisions.
- Begin by specifying what we need from this role some more so we can identify potential candidates.
- Dejan and Geoff to work on an initial list of skills.
- The board needs to discuss if there is money available.
- Summit proposed to happen in Macau on November 2nd (the day after ROSCon 2019).
- Alternative date: Morning of October 30th (before ROSCon workshops).
- A poll will be made on Discourse in the open Autoware category; all TSC members are required to answer.