Commentary about robot configurations and running multiple robots

[Raymond Sheh]

Hi All,


I hope you've had a great start to the first day of competition! I'm
really looking forward to hearing and seeing what you've all been up to.
Also many thanks to Amy, Erick, Kam, and our international volunteers
for running the competition!


Amy and Erick have brought up a concern that has been raised regarding
teams having multiple identical robots, and running one robot while the
other one(s) are being repaired.

TL;DR: The question of swapping robots is covered on pages 14-16 of the
current rulebook, under the definition of "Configuration". In
particular, the note on page 16 states:

"NOTE: This rule also covers the complete replacement of the robot
(which we consider to be equivalent to replacing so many parts that none
of the old parts are part of the robot). If the new robot is similar in
performance to the old robot prior to it becoming damaged, and otherwise
satisfies the criteria above regarding “no unfair advantage” to the
satisfaction of the Competition Lead(s), this is allowed."


Now for the lengthy discussion bit - I'd have liked to have a discussion
with all the teams who had concerns in person if I were there (and not
stuck in Trustee meetings) but I'm not so I guess this is the next best
thing. Sorry about that!


Of course like all rules, this rule has imperfections and we welcome
suggestions for improvements that we can enact in future competitions. I
did want to provide some commentary as to why this rule is the way it is
though. No rule satisfies every possible constraint and we had to make
some choices amongst them. It would be great to hear of suggestions for
further improvement that also satisfy the following constraints (or come
up with justifications for a deliberate choice of a different set of
constraints).


We want teams to be free to repair their robot and keep competing. For
example, if a robot has a worn servo or a broken wheel or a burnt out
controller, teams should be free to replace that component as long as it
satisfies the "no unfair advantage" rule. Ultimately we care about the
robot design (what we call a "configuration") - not the specific
physical parts being the same, otherwise we wouldn't allow any parts to
be swapped.

On a related note, we deliberately don't require that any replaced parts
need to be identical because we realize that teams are often
cost-constrained and may not have the funds for multiple identical sets
of components. Providing guidance regarding what is considered the same
configuration, including carve-outs (e.g., carrying a surrogate arm if
they break their arm in a way that means it can't stay on the robot)
means that teams can improvise repairs that are functionally similar and
continue competing. We have had teams who have broken wheels and needed
to splint them back together. It won't look completely identical but
it's possible to test that it behaves similarly enough for the competition.

It's difficult for such a rule to exist alongside a "can't swap your
robot" rule because in the limiting case, you could just bring enough
spare parts to build a new robot. I suppose we could say something about
"you must keep X amount of your old robot" or "you must keep this part
of the robot" but I've yet to come up with an idea for such a
restriction that doesn't end up arbitrary or impractical. For example,
is it really fair to have a "you can't swap the chassis" rule if someone
breaks the motor mount off their chassis (and has brought a spare
chassis)? If we say "you must have X% of the original robot" how do you
measure that, or draw the line in a way that has meaning?

I realize that there can be a cost argument - that teams with more
resources can just bring several robots and be at an advantage. This is
true. However, a "can't swap your robot" rule doesn't solve this. We
have had several teams bring several thousand dollars worth of spare
servos to the competition so they can swap them if they get damaged by
sand or worn out. This is clearly an advantage from funding and yet we
don't object to it. Also for RMRC, the cost of the robot tends to be
fairly small compared to other costs, such as for travel to the venue.
One could easily argue that a team that can afford to send 8 (6
students, 2 mentors) has a significant advantage over a team that can
only afford to send 3 (2 students, 1 mentor) but again this is something
we allow. If we really wanted to be fair on cost, we would impose some
kind of resource limit (in money, person-time, and people on site) but
again this becomes impractical very quickly.

If a team does bring multiple identical robots, we should check and make
sure they are identical, be it by documentation or by testing the new
robot. See page 15 under the section for "Repairs". This doesn't just
apply to teams swapping out robots, it applies whenever a robot has
undergone significant repairs. Do note that part of satisfying the
competition lead that the robot configuration does not provide an unfair
advantage may involve requiring the team to do some demonstrations of
tests that have already been performed, to ensure that performance is
likely to be similar, to within the expected variability in performance
on that test. This is going to depend a lot on the robot, the part
repaired, and the tests that have been performed.


Now one possible concern is that swapping the robot is a lot faster than
making repairs. I could think of a couple of ways of approaching this.

The first would be that we lean into this. If your robot has parts that
you expect to wear out quickly, such as wheels or servos or grippers,
this is incentive to create designs that make it easy and quick to swap
parts. Perhaps this is a way of reducing cost as compared to swapping
the whole robot.

The second is to lean away from this and to provide some kind of
'leveling' rule. Perhaps for next year, we say that any major repair
(anything that isn't swapping the battery pack, for instance) means that
the team must take a break for at least 1 slot (10 minutes). This means
that teams can't just swap back and forth between two (identical) robots.

I'm planning on doing a round of rulebook updates in the month or two
after RoboCup, while memories are still fresh, and we certainly welcome
input on these options then.


Of course, for any given mission, only one robot is allowed to run at a
time. We have had marsupial robots in the past (one robot carrying
another robot) - in fact one of the original Open Academic Robot Kit
robots (on which RMRC was originally based) was a marsupial robot, but
that isn't covered right now in the competition (although we can discuss
this later!).


As always, we welcome comments about these rules, along with suggestions
as to actionable modifications that can be made.


Cheers!

- Raymond

--
https://raymondsheh.org