V-RoHOW Roadmap Discussion (June 26th)

Maike · July 28, 2020, 11:43am

Members from the following HL league teams were present:

Bold Hearts
NUbots
ICHIRO
Starkit
SYCU-Legendary
Bit-Bots
Rhoban
CIT Brains
Ikid

Present for the Technical Committee: Ludovic Hofer, Maike Paetzel, Jacky Baltes, Reinaldo Bianchi, Reinhard Gerndt, Soroush Sadeghnejad, Wang Hao
Discussion lead: Ludovic Hofer
Note taker: Maike Paetzel

Polls that were made during the discussion were not monitored - it was possible for more than one team member per team to raise their hand in a poll and it was possible for participants from outside the HL league to raise their hand, too. So the poll results can only give a very first impression and should be interpreted with caution.

DISCUSSION BLOCK 1 / After Slide 5 - “Initially planned for 2020, but canceled”

Question: Should it be possible to re-vote on rule changes that had already been decided?
- during the competition, it should be possible to re-vote on issues if the league saw there is a mistake with the current rules
  - This is already possible with a 2/3 majority at a competition (although not officially stated in the rules?)
Proposal: require a 2/3 majority as well to vote out rules that have already been stated in the planned rule changes in the roadmap
- at least those that had been voted on already
Comment: the TC should not be bound by the league’s voting, sometimes it is good to take opportunities
- Response: it might also reflect bad on the league if there is an opportunity, but nobody in the league can actually take it

DISCUSSION BLOCK 2 / After Slide 6 “Planned for 2021”

Comment: makes more sense to keep the changes planned for 2021
POLL:
- Yes (postpone it to 2022): 11
- No (keep it for 2021): 1
Comment: some things can still be developed for 2021 (like the message filter in the GameController) and tried out during the tournament

DISCUSSION BLOCK 3 / After Slide 7 - “Discussion: Height limits”

Question: there are not too many teams that are in the ‘gap area’ in terms of height anyway - why can’t we have a gap?
- what happens to teams that might fall into that gap in the future? - do we want to exclude the ‘perfect robot’ that falls into this gap?
  - What happens if you have the perfect robot but the foot is a little too large, or the robot is a little too heavy - we are already excluding teams based on some measures, why would the height be so different?
- there used to be a gap in previous years, especially to not make teams add a very small gadget to the robot to go into the next higher class
- how likely is it that this happens?
especially for the small robots it might be difficult to play with that large robots
- the Darwins should not be the deciding factor for an entire league and there are already other constraints that are hard for the Darwins (ball size, grass length)
if we allow a dead zone, we could vote on the height limits in the leagues separately
from the perspective of a team with large robots in the KidSize league, there is already a concern of breaking the smaller robots
- and even for the teams with large robots there is a concern of playing with even alrger robots (even between 70cm and 120cm robots there is a huge difference)
similar concerns raised by different teams from KidSize
Proposal: Split vote, both on having a deadzone and then vote separately in KidSize and AdultSize

DISCUSSION BLOCK 4 / After Slide 8 - “Discussion: Next major update”

what are upcoming technical challenges? Might be interesting for teams to know so they can assign these to thesis students
- not important to know the exact year of the challenge being introduced, could also just generally be for one of the future years
- challenge doesn’t need to be completely defined, but at least roughly so it’s clear what the challenge is about
- Proposal: include upcoming technical challenges in the Roadmap
it would be nice to have more ideas for technical challenges being proposed from within the league
Next major rule update in 2023 or 2024?
- it would probably be reasonable to delay everything for a year including the next major change (there was no opposition)

DISCUSSION BLOCK 5 / After Slide 12 - “Discussion: Outdoor?”

artificial turf is fine, but only if we go for decent one. In the last years, the organizers have gone to just a better version of a carpet
nice idea to have a path to playing outdoors (for example by playing outdoor but with a roof over the field)
- natural light is already allowed, but was only used for one field - maybe could be used for all fields before going outdoor?
- might be problematic to make this a requirement for the LOC
- it is even hard for SPL to get outdoor fields
- natural light conditions in different countries might also be different, hard to practice
How much would it actually change for the walking to have real vs artificial grass?
- Robots need to be waterproof to play on real grass, because we cannot assure there is not a drop of water in the grass. And that doesn’t have anything to do with the ability to play soccer in particular
SPL and Humanoid league using the same size field, are nearly the same size, have nearly the same rules, but SPL seems further in the development of team strategy. However, they never play each other - would be nice for a comparison
- problem with surrounding parameters like field markings, goal sizes, balls that differ between leagues and would be to the advantage of one of them
- political problems what happens with the league of the ‘losing team’
Bigger field in KidSize might come in 3-4 years already according to the current events foreseen in the roadmap
- it is going to be difficult for teams to have 6 functioning robots - importance for more joint teams
- how can the TC/OC help teams to form joint teams ahead of RoboCup?
- once we have six robots per team, robustness will become one of the most important factors
  - is it really going to be robustness or rather money e.g. more people and / or more robots?
- minimum number of robots allowed that is < 6 might be good for the beginning
How do you deal with one trophy for a joint team? Medals instead?
- having another team in close proximity is important, but also networking
- foster remote collaboration might help?
- at some point you do want to have a test game with robots from both teams - not everything can be tested in simulation and/or remotely
- working remotely is possible, but it has limitations especially if hardware is involved
- Travelling is a problem, but also funding for joint teams - applying for funding that fosters collaboration?
- enforcing a joint protocol, but also enforcing a common definition of functionality like ‘go to this location’
  - high-level strategies should align when playing together
  - provide an incentive for applying the common protocols
    - eventually we will need to have joint teams - how can we have an incentive? Maybe an award or a certificate for best collaboration project
  - Technical Challenge for collaboration between two teams (blindfolded robot might work for this) - many voice support for this in the chat
  - What if one team doesn’t find a partner?
    - teams should be agreed on beforehand - but that may still be a problem for a team not finding another team to collaborate with
    - allow teams to collaborate with more than one team and they only take the best score
    - GameController could step in for a team without a partner? would need proper implementation
Field size
- AdultSize field won’t fit inside a typical football hall anymore at some point - this needs to be mentioned in the Roadmap that at some point the field size can only grow if the field moves outdoors
- might even be unreasonable to find one place that provides multiple soccer halls right next to each other - so we might need to go outside earlier
- when would corner flags be included? (at what size of the field)
  - it is already harder for localization on the AdultSize field than on KidSize
  - adds more costs to the field right now
    - response: it is actually not that expensive and would come at some point with the fields we are renting
  - is anything keeping us from using a corner flag already?
    - response: Less artificial landmarks were voted for at some point, but that was when we still had smaller fields
  - corner flags could be transported from one venue to the next - but that has transportation costs as well for the person who needs to do that
- concern is raised regarding the fact that we should focus on robot abilities rather than increasing the size of the field
  - response: One of the core element for playing soccer is bipedal locomotion which has no major incentive if we do not increase the size of the field

DISCUSSION 6 / After Slide 14 - “Gameplay rules”

Don’t mess up time (DMUP) doesn’t give an idea about how long a robot can run with one battery
Technical challenge idea: run 15 minutes without changing battery
- Problem: challenges are supposed to last less than 2 min per challenge
- you also want to test autonomy during real games where they have to perform
Change the criteria - maximal DMUP instead of average?
DMUP is not well defined enough - what is with a goalkeeper that just stands around for 10 minutes? It is still playing, technically
- just exclude the goalkeeper? There is no incentive to just have field players standing around not moving for an extended period of time
- include distance travelled by a robot during a half-time
Proposal: Increase time for semi-finals and finals
- good idea from an organizer’s perspective (tradeoff to let teams play enough games within the possible time slots, but there are more open timeslots towards the end)
- maybe for all knout-games from quarter finals onwards
- POLL: making games longer from quarter finals onwards
  - Yes: 8
  - No: 0
Question: How long does your robots actually perform? (responses were given in the chat)
- Rhoban: 30 - 40 min
- NUbots: No official metric, but we are more than comfortable with a single battery per half
- Starkit: > 20
- Bold Hearts: no official metric either, but we are usually fine with a battery per half
- Bit-Bots: 15-20 min
Question: Are the information from the application process published anywhere (the software description fields that were introduced this year for the first time)?
- TC response: No, but they will be made available
- comment: Was a good idea, even more categories to include (+2 agreements with this)
  - there were problems with uploading the PDF files - file size too big
  - session timeout is too short on the submission page as well
- encouragement to add more publications
- review process was also quite interesting - quite nice idea to involve the teams
  - would be nice to have more standards / strict guidelines both for reviewers but also the content of the Extended Abstract, how it is supposed to look like
Maike talks about the idea of developing a platform to give teams the opportunity to upload papers / datasets / open source projects throughout the year, that are than published on the HL website

Technical Challenges

Ideas
- Moving ball -> Kick from flying ball -> kick with head -> bicycle kicks
- walking on realistic grass
- robot runs 15 minutes without changing the battery
there are several options for each technical challenge to advance
- start a discussion period in the forum, then vote on them