[robocup-small] Proposal for energy budget
Patrick Dingle
pdingle at gmail.com
Tue Sep 13 18:34:37 EDT 2005
Hi everyone,
I have not been involved in robocup for over a year, but feel I have
something to offer to this conversation.
>From what I have read there are several desires of SSL teams. The desire I
would like to talk about is the desire to move towards more "real"
soccer/football games... which have much more passing and teamwork and less
individual effort. I would also like to address the concern that robots
should need not be re-engineered and rebuilt (at least in the short term,
for '06).
I have heard two main ideas for accomplishing the goal of encouraging better
team play (e.g. passing) in the SSL.
1. Larger field and/or smaller robots
2. Energy budget or speed limits
While I am a big fan of #1, I leave it to active teams to decide whether it
is reasonable to increase the field size. I would like to address #2, which
can be accomplished in the short term simply using software, and without
forcing teams to redesign robots. There are two possible implementations
that I can think of offhand:
a) Set a cap on velocities, and possibly also a cap on accelerations. This
should be trivial for teams to implement into their systems. Teams can
continue to use their existing robots. A "referee camera" could be set up on
each field, and track robot positions vs. time. This camera would exist for
the purpose of measuring robot velocities and accelerations, and checking
that they remain below legal limits. If a robot goes to fast, play could
halt and a yellow card could be issued or whatnot.
b) Simulate an energy budget. A "referee camera" would be set up similar to
proposal (a), but more complex algoritms would be used to track how much
energy each team has used. Consider that the referee camera can see every
robot and the ball throughout the entire game. From this, it can record
location history (and also calculate velocity and acceleration histories).
Given the mass of the robots and the ball (teams would be required to
weigh-in before each game), the energy required to achieve these
accelerations can be calculated. Friction could also be taken into
consideration, but this would have to be approximated with an emperical
formula (e.g. power = 0.2 * mass * sqrt(speed)). Of course, a better formula
could be derived that should closely match what really happens. Finally, the
ball speed could be calculated and similarly related to a power function.
However, the camera probably wouldn't be able to see the ball at the speed
that many teams fire at. Rotational accelerations can be ignored - the
energy required is almost always insignificant.
I would like to suggest that (a) is a simple short term method to accomplish
what many teams are asking for. Proposal (b) also does the job, but is
harder to implement. Both of these methods place artificial energy limits,
(b) being more accurate than (a). In reality, these methods are inaccurate
because they do not take into account the efficiency of a given robot
design. However, in the short term I think it is the right thing to do. In
the longer term, teams can be expected to adhere to an actual battery energy
limit (and teams will have to face the inevitability of redesigning
effecient robots).
Finally, I would like to point out that energy is not strictly the quantity
of interest. A small, thin and agile athlete needs less energy than a large,
muscular athlete to accomplish the same tasks. However, both athletes can
play a full soccer game at the same intensity because the larger athlete has
the capacity to produce much more energy (and eat a lot more before the
game). Therefore, one might reason that if an energy budget is implemented,
the energy storage that a given robot can have should be a function of its
mass (albeit, probably not linear).
Patrick Dingle
Cornell Big Red '03 '04
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.cc.gatech.edu/pipermail/robocup-small/attachments/20050913/3dd0bc49/attachment-0001.html
More information about the robocup-small
mailing list