Hello all,

IHO the current situation is that we use a 'Nao-like' model. The dimensions, motor strengths and joint restrictions were roughly based on the Nao robot, however there has been no intention to model the Nao robot, or any other specific robot, accurately. Actually, it was never officially decided to use the Nao as a base for the 3D simulation; after the highly unrealistic soccerbot model used in 2007, this was the only more realistic model that got implemented and so it was used ever since. Also, the licensing of Aldebaran only concerns the *graphical* model, which is separate from the *physical* model, which currently is also just a collection of boxes. So, I believe we are free to change our current physical model in any way we want, as long as we do not claim that we model the Nao. As a side note, to help us there, we are also welcoming any efforts to make a good-looking, license-free new graphical model to use instead of the current model.

Now, I think it would be great to implement a range of other models, such as DARwin-OP, in rcssserver3d. Clearly we want to move towards different, possibly heterogeneous models in the future, however I feel we have no clear picture in what way to get into this, and I think having several models around with different abilities and restrictions can make things more clear. I personally would like to see us at some point move away from modelling existing models and go a step further than that, but for now getting some experience with different models could guide us as to which direction such a step should go to.

Having said this, I don't think changing the model completely is a good idea in the short term, the choice would be as arbitrary as just sticking to the Nao-based model. For this year we can focus on 11vs11 with more dynamic play. To enable this, some simple changes may be enough, such as indeed a ball with some different properties, and small changes to the current model to increase the capability of developing good ball handling skills. One example we were thinking of is for instance to change the foot shape, to actually have a rounded front of the foot, which could increase the directional control of the ball. Another thing can indeed be to allow the yaw-pitch joint to have positive angles, and-or replace it by a normal yaw joint.

To address the realism concerns, I am not a fan of restricting the capabilities of the model completely, as suggested by Pengfei Duan for instance. For dynamic games having the possibility to go fast is nice. What I would suggest is to have some energy bound instead, either globally as a per-game battery capacity, or, which we prefer, a temporary bound modelling overheating/stamina/recovering. The major foundations for these are already available in the simulator; for our paper at the RC symposium[1] we implemented an energy consumption measure based on the product of torque and angular displacement. Currently, requesting the torque applied by a motor breaks multithreaded ODE, I plan to fix this in the near future.


[1] Lattarulo, V. & van Dijk, S.G., 2011. Application of the ''Alliance Algorithm'' to Energy Constrained Gait Optimization. In The 15th Annual RoboCup International Symposium. Istanbul, Turkey, pp. 393-404.

On Sat, Jul 23, 2011 at 9:43 AM, Qingyuan Zhang <miccrun@gmail.com> wrote:
Yes, you are right, changing the current model is easy, but maybe we have to
get the permission from Aldebaran. I heard from Hedayat that there have been
 some problem in the permission of using NAO model, maybe it is not that easy
to change.

On Fri, Jul 22, 2011 at 10:26 PM, <n@ai.univ-paris8.fr> wrote:
Dear Qingyuan Zhang,
Dear All,

Darwin op seems to be less interesting than NAO. The model is simply a
collection of boxes with reductions stages (via gears) with very limited
amplitudes due to motors size. Hardware access (as it is fully open) is
not really interesting. We have to plan to synthesize softwares, for a
wide range of humanoid models, closed and opened without differences.

Increasing the yaw angle max (for positive angles) on the current model
that simulates NAO, is easy to do. The current version of the simulator is
an old model of the NAO and this is different on new ones. With positive
yaw angles, we could rotate around a fixed point in front of the NAO.

Integrating different models of humanoid is also very interesting, but it
needs much work.

We have to avoid non realistic walks.
Maybe, integrating new physical foot contact is an interesting issue, that
allows to have different robots without much work. Surnatural walks are
probably linked to the exploitation of a long time trial of the
foot-ground contact model and modifying it could be interesting. It could
also be an opportunity to test the robustness of walks, that is a central
question for humanoid motions.

FCPortugal said that we could increase the ball size, that is maybe a good
and simple idea to promote real kicks during the game, promoting
multi-player soccer in the same time.

best regards
Nicolas (L3M-SIM)

10 Tips for Better Web Security
Learn 10 ways to better secure your business today. Topics covered include:
Web security, SSL, hacker attacks & Denial of Service (DoS), private keys,
security Microsoft Exchange, secure Instant Messaging, and much more.
Sserver-three-d mailing list

Storage Efficiency Calculator
This modeling tool is based on patent-pending intellectual property that
has been used successfully in hundreds of IBM storage optimization engage-
ments, worldwide.  Store less, Store more with what you own, Move data to
the right place. Try It Now! http://www.accelacomm.com/jaw/sfnl/114/51427378/
Sserver-three-d mailing list

Adaptive Systems Research Group
Department of Computer Science
University of Hertfordshire
United Kingdom