[robocup-msrs] Answers to questions

Kyle Johns kylej at microsoft.com
Fri Jun 8 14:05:41 EDT 2007


Thanks for the great questions and bug reports, Jackson.

1- The ball size will be the nowadays ball size?
It appears to be too small for the Robudogs. Or
the Robudogs appears to be too big for the ball.
And as it is very easier to have a larger ball than a smaller Robot...

We're planning to keep the ball and robots to be the same size unless this proves to be a problem.  Everyone please speak up if this turns out to be an issue.

2- Sorry for my bad English understanding in this
question, but in question 2 below:
 >>> "2. Are we allowed to create our own model
for the competition? ( e.g. dribbler , kicker ,etc)
 >>> Only the Robudog models will be allowed in the competition."
As models you mean just the robots or also the
movement sequence, as walk, turn, kick. Or asking in another way:
Will we be able to a) Define our own movement
sequence for the robots to walk faster and kick
more efficiently or b) We will be able only to
decide when our robots must do each of the predefined movements?

You must use the Robodog model but you can define your own movements.

3- About the "the robot leaves the field
boundaries" rule. Sometimes a Robot must go out
of the field to kick a ball near the line, not?
So I think that you should establish a "second
field boundary" for the Robots, at least "robot
size" greater than the field boundaries.
Also I just noticed that the attackers are using
the boundaries, i.e. the attackers are being
teleported back to the field, but the goalies may
move freely outside the play field.
And, if you use teleport (move/beam), will you
ensure that the teleported robot will not collide
with a in field robot? This happened with me a
couple of times... and as 2 bodies cannot occupy
the same space at the same time, the 2 robots
were just throw away... very funny to see, but not at a compettition :-).

Good observation.  We will need to extend the boundaries as you suggest.  We'll also do a little more work on the teleport mechanism to prevent the robots from being flung :-).

4- Finally sometimes at the start of the
simulation, some of the Robudogs just fall aside
and the ball also appears at a different position
than "center of field". I think that the initial
"teleport" routine must be checked.

More teleport work.  Thanks.

5- This is specific for he NTX Soccer Simulation,
but... with the NXTs, after a goal the robots and
the ball are continuously teleported from second to sencond.

I just recently noticed this after updating to the latest version of the Ageia engine.  I fixed it with a referee change.  Hopefully everyone is using Robudogs at this point and this won't be a problem.

-Kyle


-----Original Message-----
From: Jackson Matsuura [mailto:jackson at ita.br]
Sent: Friday, June 08, 2007 10:37 AM
To: Kyle Johns; robocup-msrs at cc.gatech.edu
Subject: Re: [robocup-msrs] Answers to questions

Hi Kyle, hi all.
I'm Prof. Jackson Matsuura, Manager of the
Brazilian National Team, a union of 12 Brazilian
Institutions that will participate as a unique National Team in Atlanta 2007.

We are qualified in the Humanoid, Small Size,
Soccer Simulation 2D, 3D and PV, Rescue
Simulation Agents and Virtual Robots and in the
Junior Dance and Rescue Leagues/Sub leagues.

About 6 students and 4 professors involved with
SoccerSim2D and 3D, Virtual Robots and with the
Humanoid development are now working partial time
with the MS-RS. We have also some other students
just working with MSRS to use it in their graduate works.

I myself am now switching my own efforts between
MSRS, Humanoid and SoccerSim3D.

We just are not sure if we will have resources
for one more competition. We still do not have
enough funds to take the team members in the 9
Leagues and Sub leagues to Atlanta. Probably one
of the members of these Leagues/Sub Leagues will
also be responsible for our team in the MS-RS Challenge.

Now the questions:

1- The ball size will be the nowadays ball size?
It appears to be too small for the Robudogs. Or
the Robudogs appears to be too big for the ball.
And as it is very easier to have a larger ball than a smaller Robot...

2- Sorry for my bad English understanding in this
question, but in question 2 below:
 >>> "2. Are we allowed to create our own model
for the competition? ( e.g. dribbler , kicker ,etc)
 >>> Only the Robudog models will be allowed in the competition."
As models you mean just the robots or also the
movement sequence, as walk, turn, kick. Or asking in another way:
Will we be able to a) Define our own movement
sequence for the robots to walk faster and kick
more efficiently or b) We will be able only to
decide when our robots must do each of the predefined movements?

2a) If the asnwer id b), What should we do when
the Robudog stuck at the middle of a movement?

3- About the "the robot leaves the field
boundaries" rule. Sometimes a Robot must go out
of the field to kick a ball near the line, not?
So I think that you should establish a "second
field boundary" for the Robots, at least "robot
size" greater than the field boundaries.
Also I just noticed that the attackers are using
the boundaries, i.e. the attackers are being
teleported back to the field, but the goalies may
move freely outside the play field.
And, if you use teleport (move/beam), will you
ensure that the teleported robot will not collide
with a in field robot? This happened with me a
couple of times... and as 2 bodies cannot occupy
the same space at the same time, the 2 robots
were just throw away... very funny to see, but not at a compettition :-).

4- Finally sometimes at the start of the
simulation, some of the Robudogs just fall aside
and the ball also appears at a different position
than "center of field". I think that the initial
"teleport" routine must be checked.

5- This is specific for he NTX Soccer Simulation,
but... with the NXTs, after a goal the robots and
the ball are continuously teleported from second to sencond.


I think that it is all for now. And we are
thinking in reproduce the MS-RS Challenge in the
RoboCup Brazil Open this year (October at
Florianópolis - Brazil). Who should we contact to
have permission and perhaps some support to do this?

Thanks for the attention

Jackson Matsuura
Brazilian National Team Manager
Brazilian National Team Site:
http://natalnet.dca.ufrn.br/wiki/doku.php?id=robocupbrazil:brazilian_robotic_team
RoboCup Brazil Committee: http://robocup.sorocaba.unesp.br/
RoboCup Brazil Open: http://robocup.sorocaba.unesp.br/brazilopen2007/


At 20:32 7/6/2007, Kyle Johns wrote:
>A couple of you have introduced yourselves and
>asked a few questions.  Here are my answers to
>those questions and I'm still looking forward to hearing from more of you!
>
>-Kyle
>
>1. Which model will be used for the competition? ( lego or robudog)
>
>The Robudog models will be used for the competition.
>
>2. Are we allowed to create our own model for
>the competition? ( e.g. dribbler , kicker ,etc)
>
>Only the Robudog models will be allowed in the competition.
>
>3. I noticed my control algorithm is affected by
>the frame rate of the system. is there any simple method to solve this problem?
>
>The best thing to do here is to find the fastest
>machine with the best graphics card that you can
>for now.  At Robocup, the simulator will be
>running at a fairly constant 58 frames per
>second.  I will do some more work in this area
>to give you some hints on how to debug your
>soccer player services at lower framerates.
>
>4. There is no fouls mentions in the document.
>Is it allowed to push the opponent robot when getting the ball?
>
>I believe we'll adopt the four-legged league
>rule about bumping the opponent.  Any contact
>with an opponent that lasts longer than three
>seconds will be a foul and the offending robot will be penalized.
>
>1) How shall a robot know the referee port to
>connect to? (I guess we can get some help on  that in Atlanta)
>
>We'll give you the IP address of the referee
>machine in Atlanta.  It will be a very simple
>process to update your manifests to connect with the referee.
>
>2) How many players per team? 1 goalie + 1 attacker?
>
>That's right.
>
>3) What is meant exactly by "the robot leaves the field boundaries"?
>
>Each robot is identified by a single point which
>is its origin.  When that origin crosses outside
>of the white lines at the field boundary, the
>robot will be penalized.  This is enforced
>automatically by the referee service.
>
>4) Are there any constraints about the positioning/motion of the goalie?
>
>There are no constraints on the positioning or
>motion of the goalie other than that he must
>stay inside the field boundaries.  The goalie
>can go inside of the goal while the offensive player cannot.
>
>5) While watching a game, how can we arbitrarily
>move to a different viewpoint outside the field
>(to have a panoramic view of what is going on), how can we zoom in, etc?
>
>You can move the camera by pressing the left
>mouse button and dragging the mouse across the
>simulation window.  Pressing the W key moves you
>forward while pressing the S key moves you
>backward.  The Q key moves you upward while the E key moves you downward.
>
>6) Is the software final? You mentioned in a
>previous email that Microsoft will ship an
>update to the soccer package "in about two
>weeks", will this happen? I hope not, but if yes what should we expect?
>
>We have decided to keep the software as it is
>for now.  We will be using an updated referee
>service in Atlanta but we'll only be adding some
>user interface to it to make it easier to
>control the match.  We will not be changing the
>referee behavior and we don't anticipate that
>any other Microsoft software will change unless
>we find a serious bug.  Likewise, I expect that
>the Robosoft software will not change unless there is a required bug fix.
>
>7) Is it possible to get ground-truth
>information about the game state (robots/ball
>position) during the development phase?
>
>Since we won't be shipping an updated referee
>service, this is a little bit difficult.  There
>is a way for you to query for your robot's
>position, though, and I'll provide instructions in a follow-up mail.
>
>8) How can we reset robots/ball to specific positions during development?
>Moving the ball to a specific position is easy,
>moving the robots is more difficult.  To move
>the ball position, put the simulator into edit
>mode by pressing F5.  This pauses the physics
>engine so your soccer players and the ball will
>stop moving.  Find the Soccer Ball entity in the
>upper-left-hand pane and select it.  In the
>lower-left-hand pane, select the position
>propery.  Either type in new position numbers or
>press the control key and drag the ball around
>the field with the left mouse button to move
>it.  When you have it positioned properly, press
>F5 again to return to simulation mode.
>I'm still working on an easy way to move the
>Robudogs around and I'll let you know as soon as I have it.
>
>9) How can we guarantee that testing on a slow
>machine and running on a fast machine in Atlanta
>will not produce different results? What are the
>MSRS primitives that allow a universal (CPU
>speed independent) measurement of time?
>
>The measurement of time is not really the
>problem, it is the accuracy of the physics
>engine which is affected by having a slower
>machine with a slower framerate.  The referee
>machine in Atlanta will be capable of
>maintaining a framerate of 58 frames per
>second.  The MSRS environment is essentially
>asynchronous outside of the simulation engine so
>your service will receive camera frames
>approximately every 350 ms unless the simulator
>framerate drops below 3-4 frames per second.  If
>your simulator is running that slowly, you need to find a faster machine.
>Another option is to replicate the setup we will
>have in Atlanta by using three machines instead
>of one.  Use the machine with the fastest CPU
>and the best graphics card for the referee
>machine and two other machines for each of the
>teams.  I'll send a manifest for this configuration soon.
>_______________________________________________
>robocup-msrs mailing list
>robocup-msrs at cc.gatech.edu
>https://lists.cc.gatech.edu/mailman/listinfo/robocup-msrs




More information about the robocup-msrs mailing list