Tinker Time

[Alistair Lynn]

Hi all–

It may be worth taking a bit of time to consider whether tinker time is still useful in the competition.

For those unfamiliar, tinker time refers to the first morning of the competition, in which teams can work on their robots at the venue. There is some kind of system for allowing teams to book time in the arena so they can test in a live environment.

There are a couple of problems with this system:

1. With the number of teams we now have, they get very little time in the arena;
2. For teams who are ready to go, it’s a waste of a morning;
3. (Arguably) teams will get more done once they actually see how their robot has performed in a live environment–that is, between matches.

Particularly for former (or current?) competitors, what are your thoughts on the usefulness of tinker time? More generally, is it beneficial to scrap it and instead take the freed-up time to have a longer league, or bigger spaces between matches?

Discussion time: 2 weeks

Alistair

[Murray Colpman]

I think it's a trade off. I think lots of teams do come to the competition having never tested their robot in an actual arena, or even having not finished their robot. While this remains true, tinker time in some degree I think OS a necessity. It also allows teams time to settle in, especially late arrivals and in case we blueshirts are late setting things up (see last year) or there are problems (eg the WiFi). I think for these reasons we should probably keep it in some capacity, though I admit these aren't very good reasons.

Murray.

--

---
You received this message because you are subscribed to the Google Groups "Student Robotics" group.
To unsubscribe from this group and stop receiving emails from it, send an email to srobo+un...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Andrew Barrett-Sprot]

Personally, I think most teams both need and expect time to check that their robot still works on the day of the event, after producing horribly panicked code the night before. We already have plenty of non-moving robots in the first round, and eliminating tinker time will only make it worse.

Also, +1 on Murrays statement, It's very hard to recreate the arena for teams (esp. lighting conditions), so having time to tweak the robot before the event is very important.

On the other hand, I remember the time between matches being uncomfortably small when I was competing, and anything that increases that time would allow for more major modifications to the robots. 

[Peter Law]

Hi,

When this was last discussed in IRC, one of the ideas which popped up
was that we really ought to have testing arenas or part arenas at the
venue. If there's space this is something I'd be massively in favour
of anyway.

Competitors: would that compensate for removing tinker time?

Murray wrote:
>> teams settle in ... and ...

>> in case we blueshirts are late setting things up (see last
>> year) or there are problems (eg the WiFi).

I'd much rather see us just not put any matches during, say, the first
hour on the Saturday as a buffer for this sort of thing. If we're not
managing tinker time (and there are testing areas instead) then this
frees up Blueshirts to be fixing things

I'd guess that removing tinker time it also ensures that everyone has
had a fair start (rather than only some people getting in the arena).
Do we have data on whether:
a) everyone got in the arena during tinker time?
b) everyone _who wanted to_ got in the arena during tinker time?

Andy Barrett-Sprot wrote:
> Personally, I think most teams both need and expect time to check that their
> robot still works on the day of the event, after producing horribly panicked
> code the night before. We already have plenty of non-moving robots in the
> first round, and eliminating tinker time will only make it worse.

Surely non-moving robots in the first round is pretty much a
certainty? I think what we need to capture is the number of teams for
whom tinker time makes the difference between wasting the first, say,
2 matches and not wasting them.

> Also, +1 on Murrays statement, It's very hard to recreate the arena for
> teams (esp. lighting conditions), so having time to tweak the robot before
> the event is very important.

This is something we're trying to get better at, though I'm a little
surprised that lighting remains an issue -- in my experience libkoki
is pretty good at seeing things now. Do we have any data on how
differently it performs in an arena than a well-lit classroom?

> On the other hand, I remember the time between matches being uncomfortably
> small when I was competing, and anything that increases that time would
> allow for more major modifications to the robots.

Would you prefer to have two matches close together and then a big gap instead?

Thanks,
Peter

[Harry Cutts]

Hi,

On 2 February 2015 at 17:19, Alistair Lynn <al...@studentrobotics.org> wrote:

It may be worth taking a bit of time to consider whether tinker time is still useful in the competition.

During our little competition at the active day on Saturday, I was really struck by how important the time between matches is. It was really difficult to get large things done during even a 1-hour gap. So, I'm certainly in favour of increasing gaps between matches at the expense of tinker time. Test arenas could help us to reduce the need for it, although there's still the issue of demand greater than supply. Maybe we could ask that only rookie teams use tinker time in the arena (or just prioritise them)?

Harry Cutts

[Tom Leese]

Hi Alistair,

On 02/02/15 17:19, Alistair Lynn wrote:
> It may be worth taking a bit of time to consider whether tinker time is still useful in the competition.

Based loosely on what I wrote on IRC.

I think we can afford to shrink tinker time down to the minimum length
required so that all teams get one chance in the arena, and then
increase the time between matches.

When I was a competitor, we made more useful changes in the gaps between
matches because we had a quicker turn around between making the changes
and seeing how they fare in the game. Tinker time was not that useful to
us and we mainly just ended up making minor cosmetic changes that didn't
really improve our robot.

With adequate communication, we can also make sure the problem of teams
arriving expecting tinker time to be solved and so all the teams should
come prepared.

Tom

[Lila Fisch]

Hi,

just my 2 pence from sigining teams into tinker time slots:

Just getting a slot and coming on time for it takes up large amounts of time and discussion for teams.

That could be reduced by assigning slots though.

cheers

lilafisch

--

--- You received this message because you are subscribed to the Google Groups "Student Robotics" group.

To unsubscribe from this group and stop receiving emails from it, send an email to srobo+unsubscribe@googlegroups.com.

[Howard Buck]

> On 2 Feb 2015, at 21:33, Harry Cutts <eterna...@gmail.com> wrote:
>
> It was really difficult to get large things done during even a 1-hour gap. So, I'm certainly in favour of increasing gaps between matches at the expense of tinker time

I'm not sure this needs to be the case. A good development strategy would be to keep a stable solution on one memory stick for matches, and develop on another stick in between. Once the "large" change is stable just swap the sticks and start on the next change. (Pulling the robot out of development for a match shouldn't be a hassle, and gives the non-coding team members stuff to do, and scores them points, and gives spectators something to watch, and is fun, and is the point in being here.)

Many, frequent matches are really good for getting quick feedback, if anything I would try and squeeze more in ;)

Maybe tinker time isn't ideal. It would be better to have many mock arenas all over the place, but this adds cost and logistics which is unlikely to happen for SR2015. Perhaps we could just start the league straight away and say that the first two matches a robot competes in don't score points? Practice rounds, if you like.

We should definitely start the (real) league later on the Saturday than the Sunday, to allow people to get their bearings, get wristbanded, unpacked, and for blueshirts to deal with any unforeseen circumstances.

Howard

[Lila Fisch]

Hi,

I'm not sure this needs to be the case. A good development strategy would be to keep a stable solution on one memory stick for matches, and develop on another stick in between. Once the "large" change is stable just swap the sticks and start on the next change.

The hardest bit about this is getting teams to do it [1].

I think we should follow the example of Germany: Have mini matches at the last Tech Day before the competition. This way teams are already put into the stress state of the competition and can practice strategies like this.

It would be better to have many mock arenas all over the place, but this adds cost and logistics which is unlikely to happen for SR2015.

I think we can do basic improvements with very little cost. I think most teams don't need a full scale, similar lighting arena to fix their basic problems.

The classic issues I encountered:

- We don't even have markers > put up some markers along a wall

- We don't have a flag (token) > make some additional flags. Cutting flags at uni should lower cost significantly enough to be able to afford some additional ones.

- The competition mode doesn't work [2]! > nope. team is panicking and probably moved the camera before going in arena. To stop them from getting stuck on this issue, clearly marked competition markers taped to a wall far from the arena [3] should help them believe the competition mode does work and get them to debug other issues.

> Personally, I think most teams both need and expect time to check that their
> robot still works on the day of the event, after producing horribly panicked
> code the night before. We already have plenty of non-moving robots in the
> first round, and eliminating tinker time will only make it worse.

Surely non-moving robots in the first round is pretty much a
certainty? I think what we need to capture is the number of teams for
whom tinker time makes the difference between wasting the first, say,
2 matches and not wasting them.

In my experience teams are still optimistic during tinker time because they still mostly do partial tests and not a proper competition run. They then get a shock when the robot doesn't work in the first match and start working on their issues. By Sunday morning many robots are doing good.

Maybe we need to get to the shock of the robot not working earlier in the day. We could just split tinker time in two?

cheers

lilafisch

[1] I have spent good parts of Saturday trying to get a team to implement this at sr2013. In panic they still kept getting confused which was which stick, despite the use of red tape on the competition one.

[2] At least I have spoken to several teams with this problem in the last two years. Not being able to disproof this issue makes it hard to get them to think about something else. Last year I sat down with a German team for a while and verified that their camera was aimed too high, using a hand drawn competition marker.

[3] In Newbury we could use a wall on the top floor and add red text: 'Competition Marker, do not move!'

[Jeremy Morse]

Hi,

On 02/02/15 17:19, Alistair Lynn wrote:

> 1. With the number of teams we now have, they get very little time
> in the arena;

This is only changed by scaling the number of teams or the number of
arenas. Any fair allocation will always give people little time in an
arena. There are 16 hours in the weekend between 9 to 5, making 32 arena
hours, divided over 54 teams. They're never going to get more than 40
minutes over the whole event [0].

> 2. For teams who are ready to go, it’s a waste of a morning;

For teams who aren't, it's vital. A hypothetical perfectly prepared team
will also be wasting the 25 / 30 minutes between matches too, which we
can't and shouldn't resolve.

It's not like that period is completely idle either: un-rushed safety
checks of all robots need to take place, competitors need to know how to
operate their robot in the arena (corner-specific USB sticks this year,
competition mode etc), and we have breathing space for all the things
that inevitably go wrong.

> 3. (Arguably) teams will get more done once they actually see how
> their robot has performed in a live environment–that is, between
> matches.

They get the opportunity to test that during tinker time anyway.

I like Lilafisch's idea of pre-scheduling tinker time in arenas: this
reduces logistics and gives people a taste of the live environment while
having good opportunity resolve any oversights they've had.

Increasing time between matches punishes the well prepared in favour of
the ill prepared; the trade off isn't really worth it. Better would be
to advertise better to teams that they can skip/forfeit matches: then
the ill prepared will get a greater gap for patching things up and will
be penalized accordingly. This gives everyone a score proportionate to
their actions/ability.

[0] SR2014 had sufficient dropouts that a 48-team schedule was run.

--
Thanks,
Jeremy

[Jeremy Morse]

Hi,

On 03/02/15 12:31, Jeremy Morse wrote:
> It's not like that period is completely idle either: un-rushed safety
> checks of all robots need to take place, competitors need to know how to
> operate their robot in the arena (corner-specific USB sticks this year,
> competition mode etc), and we have breathing space for all the things
> that inevitably go wrong.

Clarity: "that period" being the tinker time period, not the gap between
matches.

--
Thanks,
Jeremy

[Lila Fisch]

Hi,

>  1. With the number of teams we now have, they get very little time
> in the arena;

This is only changed by scaling the number of teams or the number of
arenas.

Last year we only had one arena in tinker time. Let's try our best to have two this year.

 

>  2. For teams who are ready to go, it’s a waste of a morning;

For teams who aren't, it's vital. A hypothetical perfectly prepared team
will also be wasting the 25 / 30 minutes between matches too,

or they will watch other matches to plan strategies :o)

 

It's not like that period is completely idle either: un-rushed safety

checks of all robots need to take place, ... and we have breathing space for all the things
that inevitably go wrong.

I like Lilafisch's idea of pre-scheduling tinker time in arenas: this
reduces logistics and gives people a taste of the live environment while
having good opportunity resolve any oversights they've had.

Safety checks are a very good point. And maybe we can combine things:

Have the safety checks scheduled as well;

I don't remember checking for safety before letting teams in the arena during tinker time. We could check teams before they go into the arena, right next to the arena. Teams that fail can still test but need to make changes before competing. We get most teams coming to us and don't have to chase them ourselves; this should reduce the number of people we need for safety checks. We could use those freed up people for fixing whatever goes wrong.

cheers
lilafisch

[Alistair Lynn]

Hi all–

> Discussion time: 2 weeks

So I want to conclude this relatively soon.

The general conclusion seems to be that tinker time is still necessary (to give us some margins on getting things ready, and to help teams who haven’t done anything) but:

(1) doesn’t need to be massively long, and
(2) should have pre-arranged slots for teams to tinker.

Is that about right?

Alistair