[robocup-nao] SPL Rules, especially Pushing

[Stefan Czarnetzki]

Hi Michael,

the video idea is very good! I'll put some scenes together when I find 
the time...

Until then, 2 scenarios in this video for example:
http://www.youtube.com/watch?v=58jIeJ_kPT0
(Don't say it: Yes, my students got confused with our two cameras and 
uploaded this with a wrong aspect ratio.)

A robot walking clearly towards the ball touches a standing robot. Even 
if the standing robot has no clue about the ball or where it is, the 
other one will be penalized for pushing. This really prevents fluid game 
play!
The scene at around 7min 25sec is about the closest I could find just 
now, but here the blue Nao sees the ball shortly after the "pushing". 
This by the way is a good example of a formally correct pushing call 
that might have been avoided for nicer game play. Both robots were still 
standing and the opposing robot was heading straight for the ball 
without any further obstacle, since the other robot went away.
9:20 is similar, but the other way around.

In other situations a ball came to lie very close to the goalie without 
the goalie reacting in any way. Nothing prevents a goalie from 
stretching its arms as wide as possible. If a ball lies below the arm, 
it is not inside the convex hull of its lower parts (ie. legs) and there 
is no ball holding situation. On the other hand, the goalie cannot be 
called for pushing if it looks at the ball or even if it simply does 
nothing and keeps standing there. Every player going somehow to the ball 
will eventually have to touch the goalie, which then will be pushing. 
The situation around 3:40 is close to that, but here it didn't come that 
far since the attacker didn't see the ball and global game stuck was 
called. But other goalies were blocking more space with their arms and 
similar situations occurred. I guess this could also happen with 
stationary field players and the current rules.


Cheers,
Stefan


Michael Quinlan wrote:
> Hi Stefan,
>  This exact discussion is also happening on the TC, as it was evident 
> to all that the German and US opens were refereed differently. I think 
> the true answer lies somewhere in the middle.
>
>  In hindsight we probably allowed a little too much pushing at the US 
> Open, in particular some pushing in the back that should have been 
> called. However, I agree that we should not be pre-empting pushing 
> call. One key difference was at the US Open, if it was not clear who 
> was pushing and the contact was reasonably light then we let both 
> robots stay on the field even if they both eventually fell over, I 
> prefer this then penalising both robots.
>
>   We are trying to formailse the interpretation of the rules now, so 
> we are open to suggestions (like this email) on the direction we 
> should go. I've also thought about putting together a video on 
> different pushing scenarios,  and how they should be called, it may be 
> easier to get league-wide consistent interpretation from a video then 
> from text.
>
> Michael
>
> Stefan Czarnetzki wrote:
>> Hi everyone!
>>
>> After the German Open and the US Open we all got a good impression 
>> what games will look like this year in Graz, and I think it's safe to 
>> say that there won't be any finals or semifinals without Naos 
>> touching the ball or doing anything useful at all. Nice progress!   :)
>>
>> Anyway, another point that was interesting to see and to compare was 
>> the application of the rules to the games. On the German Open the 
>> rules were applied to the word and the referees were interfering a 
>> lot, especially concerning the pushing rule; which was absolutely 
>> correct and fair. Besides this, referees were still very worried 
>> about the robots and were running onto the field in lots of 
>> situations when a robot were in danger to fall over.
>>
>> On the US Open the referees seemed to stay away from the carpet most 
>> of the times and robots were removed only in some rare situations. 
>> The games therefore looked a bit more like Aibo games, just in slow 
>> motion. If robots were close together and touching every once in a 
>> while, they were left alone. If a robot fell down, it was allowed to 
>> stand up on its own without someone running to help. The Nao that was 
>> still standing had its chance to kick during that time without 
>> interference.
>> At least this is my impression from the videos, I haven't been on 
>> site myself.
>>
>> We had intended to discuss the rules again after the big Opens and I 
>> would like to hear other thoughts, especially about the pushing rule.
>>
>> In my opinion the Naos are robust enough to survive falling down 
>> several times (as Aldebaran said), and with nice looking games 
>> without too many interventions in mind, I vote to soften the pushing 
>> rule.
>> I acknowledge that it would be much more sophisticated soccer playing 
>> if all robots would smoothly evade each other without collisions 
>> (which I haven't seen so far from any team), but I prefer the US Open 
>> style interpretation to the German Open one, where robots were taken 
>> out when heading straight for the ball but failing to avoid others 
>> correctly.
>>
>>
>> Cheers,
>> Stefan
>>
>>   
>
>