Real vs. Theory, Week 0.5

[Chak]

Originally Posted by msigalov

Doesn't only one breach in auto count? I thought I read something in the rules about that.

Only the first crossing by a robot in auto counts for points. An alliance of 3 robots can get 30 points by having all 3 robots cross a defense in auto.

[PayneTrain]

Originally Posted by Donut

This surprised me, I would have thought for sure we would see more low goal capable bots given how important capturing is for the playoffs. There were ~100 more high goals than low goals. Sounds like everyone went high if they were bothering with scoring boulders at all, and since most shooters probably can't be adjusted well for scoring in the low goal those that were ineffective just didn't score. Saw a few matches today where an alliance would breach and then not have much else to do until the end game.

It's a confluence of factors:

-Most teams that designed for low goal actually can't score in the low goal (they can't intake a ball unless it is at a dead stop right in front of them and/or they forgot how that pesky batter makes it more of a challenge)
-Teams that designed for the high goal may have designed their robot to not low goal after it's brought into their high goal scoring mechanism. I wouldn't let Jesus take the wheel on that one, but teams decided they could live with it.
-Willful ignorance of how captures work and/or drive teams at Palmetto conspiring to hospitalize me by forcing me to watch them leave rank points on the board to do silly things. It's Week 0.5 so only a few teams actually have bothered to make an educated guess at the meta and design and coach day-by-day around its development. It probably won't be uncommon to run into a drive team at early events that thinks they're really sticking it to the other teams at the event dropping the opponent tower to 0, -1, etc., while forgoing an easier 5 points that would then help transfer into an RP.

Originally Posted by marshall

The defenses are brutal on drivetrains and I'm not of the opinion that the wooden team versions provide the same difficulty that the actual ones provide. We managed to crack some versa hubs (https://twitter.com/FRC900/status/703693490752978945). Thanks to 1296 for graciously providing us with some aluminum replacements. Team 900 loves #TeamIFI!

Matches ran WAY behind schedule and I really think the defense selection process will be changed before too long to help the match reset times. This field is quite the process for volunteers too.

You weren't using those hubs as the primary support and capture point in a live axle environment were you? That would make JVN sad We killed a lot of those alongside a large chunk of my soul last year, then I remembered VEXpro has a neat website that tells us to do and not do all kinds of things with their products, like throw them on a mecanum drive wheel and think you are an expert engineer (I'm an accounting major).

I hope/expect that FIRST conducts some sort of pow-wow with head referees weekly (chain email? conference call? I'm not inclined to care how) to make sure at the driver's meeting the teams know
-G5 G5 G5 G5 G5 G5 G5 G5 G5 G5
-How defense selection works
On our team, we will have our own defense coordinator. The student build lead will transition into a role that includes selecting and communicating the selections to the drive coach.

I think I am going to break my personal best for needlessly long post but I guess I'll shoot my mouth off and people can judge me on my observations. I considered this to actually be a pretty above average event for its place on the calendar; I'd say about 1/4 of the teams here would definitely meet the (ever-declining, not their fault) bar of a championship caliber program. Some regression should be expected going into week 1. Here we go:

-I did not expect the Moat to not only have fewer selections than the rampart, but also fewer successful traversals. I guess that teams were not gunning it appropriately over the moat and turned that into a tragedy of immobility a few times, but the rampart is almost always guaranteed to be annoying. Every other series of selections makes sense. I'd say the breach count was about on par with expectations.

-While teams still have to get used to the field, they are going to pick low defenses as much as possible (audience selections and busted CDFs aside) to handle a level of visibility that is probably a half step above 2010. That plays into A+C class defense selection and its relative success rates being what they are. I don't know if the stats are available but I imagine the drawbridge and portcullis will get less and less frequent by the end of day 1 then surge back by the end of day 2. Rock Wall vs Rough Terrain seems like an observation teams easily made on the fly.

-Breach and Capture objectives were completed about as expected. I envisioned a success rate of breaches maybe 3-4% lower than observed. We'll obviously never hit 100% but Dean help you if you're a team at champs that bombs an alliance's chance at a breach and effectively crash their rankings.

-I honestly can't peg how defense will progress throughout the season, but seeing it in qualifications except under specific circumstances would be perplexing. Quals and Playoffs still have the same objective they do every year: rank high and beat everyone else, respectively. You go into every match wanting 4 RPs. You go into every playoff round wanting to stop the opposition from completing just one 5-point objective to snuff the light out of a potential 20 or 25 point bonus... right?

-Challenge vs Scale vs Sit on the Field Looking Sad Because You Left 5 Points on the Board panned out pretty close to expectations as well. LF pointed out how low Week Zero C/S rates were, but it's easy to chalk that up to teams a)not even on the field and b)tuning everything else but going up a ramp. However as we noticed in some failed attempts, maybe teams should really test that whole going up a ramp thing.

-G18 as it is can be pretty weak. Some climbers and other things were obviously out 15 inches+. On-field bumper enforcement is about as light as usual, that war is fought in the pits. I didn't personally see any missed scoring, so that's nice. Looking Forward to the TU, Week 1 events, and to get more time in the shop... we took a couple days off and I'm still antsy from that.

[Ginger Power]

Theory: Low bar robots are a bad thing for Stronghold.

I'd argue that the competitive floor of Palmetto was no lower than it has been in previous years. The upper end included a number of very good low bar teams including the winning alliance captain 179.

Fact: Defense is back in a big (Good) way.

There was negligible defense played during quals because of all the incentives to breach and capture every match. Defense showed up at the higher levels i.e. playoffs, but it didn't detract from the excitement of the game, it enhanced it. Watching teams beat a defender to score was much more entertaining than seeing a robot hit every high goal, and taking their sweet time to do so. I think this dynamic will only expand as Stronghold evolves.

Theory: Low goals are automatic, and far easier for most teams than high goal.

Low goals proved to be exceedingly difficult for teams who slowly reversed their intake to score them. The boulder had a tendency to roll down the batter. It seemed to be less of an issue for the teams that shot the ball into the low goal, but that seemed rare. I think as Stronghold develops low goals will fade out and high goals will become the more efficient alternative for the top 10% of teams. The bottom 90% would still benefit from doing low goal.

Fact: High Goals are hard.

We knew they would be. Small goal, big ball. Pretty simple math. A large majority of teams shouldn't do high goal ever. A missed shot is a capture killer in many cases.

Fiction: The defenses are hard to cross.

Most teams were able to cross most defenses under ideal conditions. Obviously ideal conditions are hard to come by when you have three opposing robots ready to disrupt you at any time, but more often than not the disincentive for defense provided room for teams to cross defenses. I think most teams would benefit from getting a larger running start for the defenses. It seemed like a large majority were trying to slowly drive up the defenses. If you're not getting air, you're not going fast enough.

Fact: The seed of your alliance doesn't matter as much as it did for that 2015 game, whatever it was called.

In 2015 it was very easy to tell which teams were going to win an event. Just add OPR's and see who has the highest. This year, an 8 can beat a 1, and it's purely based on scouting and strategy. Stronghold is the most strategically deep game since 2013, and probably since long before then. Don't get me wrong, when the top 2 teams at an event pair up, they're inevitably going to be the favorite. I will side with Car Nack on this one and say that a non first seed will win the event for more often than a typical year.

Theory: Fouls will ruin the early weeks in a similar way to how they did in 2014.

Fouls changing the outcome of a match were a very rare thing during week .5. That could be because a lot of fouls weren't called that should've been, but at least that way it's less noticable. We'll see where things go from here with regards to the outerworks being a safe zone or not. If they're not then Stronghold is going to take a step in the wrong direction.

Fact: Climbing is rare, and it gives you a nice bonus.

Climbing will become so much more valuable when breaching and capturing becomes the norm for the playoffs (I'd wager by week 4 of 5). A climb on red cancels out 3 challenges by blue, that gives huge, match-swaying power to one robot. We didn't see climbing factor into the alliance selections so much at Palmetto, but I suspect that will change when there is more parity between the top breacher/shooters, making climbing the deciding factor.

Fiction: Having a high release point is a significant advantage.

It didn't seem that this was the case from the livestream, but obviously I wasn't out there coaching or driving. Most defensive robots seemed to play drivetrain interference, rather than blocking and the blocking that did occur didn't seem to be all that effective. I didn't see one deflected shot from a blocker, just missed shots from drivetrain interference.

Fact: Stronghold is going to be epic.

'Nuff said

[Koko Ed]

Originally Posted by CalTran

You'd be surprised how many teams don't read the Tournament section of the rules.

You'd be surprised how many teams don't read the rules period.

[Doug Frisk]

Originally Posted by marshall

Matches ran WAY behind schedule and I really think the defense selection process will be changed before too long to help the match reset times. This field is quite the process for volunteers too.

I've got more but I need to get some sleep.

From my observation, the long cycle time was due to the large number of field faults ands associated replays. field reset seemed to do a pretty good job and weren't the delay.

Notes I have for Northern Lights this week include suggesting to the FTA or field manager that they walk the defenses before each match to verify they are all locked in place.

Also to make sure we get a reinforced Cheval de Friesen.

That said, in Palmetto they did eight matches per team across 64 teams. If we go with 8 per team across our sixty we should have an easier time of it. If we go with 9 matches per team things will have to go right.

[philso]

Originally Posted by Ginger Power

Fact: High Goals are hard.

We knew they would be. Small goal, big ball. Pretty simple math. A large majority of teams shouldn't do high goal ever. A missed shot is a capture killer in many cases.

^
|
Yes.

Of 84 shots on the HIGH GOAL, only 47 (56%) went in.

I logged hits and misses that I could see on the screen or I could hear the announcer commenting on beginning at about the first match in the second round of the QF's. I am sure there are some that the announcer or I missed.

Of the misses, most were high/low or off to the side. One went in one side and out the other (a heartbreaker, I am sure).

My impression is that 20% of the misses, or less, were due to defensive action. I did not notice any that were outright blocked by a defender. For at least half of the shots, the robot was unhindered by defense and in some cases could be seen taking a few seconds to adjust its aim from side to side.

My guess is that at least half of the shots that I logged were taken by teams with cameras and LED rings installed. I do not know if they used vision tracking or were viewing the image on their driver stations.

There did not appear to be any robot that was able to score in the High Goal over 80% of the time. I noticed several instances where 3824 or 179 would score in the HIGH GOAL 2 or 3 time in a row then miss about as many. This is not to criticize these two teams performance, they were probably two of the best at scoring in the HIGH GOAL at this even.


It was a bit heartbreaking to see robots slide off the BATTER after power to their robot was cut at the end of the match.

It was also sad to see robots slowly descending below the top of the LOW GOAL after the end of the match or to see the robot stop with the front bumper above the top of the LOW GOAL but the rear appear to be a few inches below.

When I had the LiveStream on in a corner at work on Friday, I recall quite a few matches where YELLOW CARDS were handed out to teams for stepping over the DEFENSES. I did not track these to see if any became RED CARDS and what the consequences were. Since this does not occur during a match, does the team lose the points from the match that just ended?

[Boltman]

Originally Posted by msigalov

Doesn't only one breach in auto count? I thought I read something in the rules about that.

RP (and 40 or 50 points in tele or more in auto)

[ratdude747]

Originally Posted by Koko Ed

You'd be surprised how many teams don't read the rules period.

*shakes head* Sadly true indeed.

Originally Posted by philso

It was a bit heartbreaking to see robots slide off the BATTER after power to their robot was cut at the end of the match.

It seems a lot of people overlooked this... I wouldn't be surprised to see a bunch of non-hanging teams retrofit braking systems.

Originally Posted by philso

When I had the LiveStream on in a corner at work on Friday, I recall quite a few matches where YELLOW CARDS were handed out to teams for stepping over the DEFENSES. I did not track these to see if any became RED CARDS and what the consequences were. Since this does not occur during a match, does the team lose the points from the match that just ended?

If it is like years previous, yes. Scorekeepers (like myself) can and will edit scores and add cards to previous matches when directed to do so by the head referee.

[Drakxii]

Originally Posted by philso

Of 84 shots on the HIGH GOAL, only 47 (56%) went in.

In my opinion for a regional that started a day after bag day that percentage is amazing. I was expecting more like 30% but I was also expecting more teams to try. So this percent might just be high given that there was less then 1 shot per match.

Also does this include auto?

[XaulZan11]

Originally Posted by Drakxii

In my opinion for a regional that started a day after bag day that percentage is amazing. I was expecting more like 30% but I was also expecting more teams to try. So this percent might just be high given that there was less then 1 shot per match.

Also does this include auto?

This was only the shooting percentage of the top 24 teams after they had 2+ days of practice. If the numbers included the entire event, I suspect it would be around 30%.

[evanperryg]

Some of my observations from Palmetto, along with some speculation on the future of this game:

I think I discussed my thoughts on breach specialists in a post a while ago. As 1296 proved very clearly, the breaching specialist can get itself through quals very easily, at least at low levels of play (I would expect this trend to continue at stronger events). They allow alliance partners to focus on other objectives, while reliably picking up an extra RP for the alliance. However, come time for elims, the breacher becomes extremely reliant on alliance partners to pick up the tower points they aren't picking up. If the breacher's alliance can't match the scoring power of the breacher, then the alliance is dead in the water.

The geometry of low bar robots makes the C defenses very hard, but a taller robot has the potential to do them much more easily. With the rarity of robots that can effectively solo the C defenses, I'd expect them to be a unique commodity at higher levels of play, where the huge amount of time they'll save their partners could be the difference between a winning and losing alliance.

I was very surprised by the amount of scaling I saw, even if few of the scalers were consistent. Scaling comes with only 10 more points than challenging, and requires some clever engineering, but clearly it's very possible, and not too difficult for taller robots. One scaler really isn't much, but an alliance with two or three consistent scalers has the ability to swing the score in their favor right at the end of a match, with no counterplay available to the opposing alliance.

This might have been my impression of the game influencing what I actually saw, but whenever the portcullis and/or drawbridge were on the field, some robots seemed more confused than normal. I'm curious to see how many lowbar teams put bendy flags or lights sticking up out of their robot just so they can see themselves.

I was very, very impressed by teams 179 and 2393. I don't know exactly how 2393 works, but I'm guessing that, based on the green light ring, they are using vision tracking. With good code, and a good shooter, it is completely viable to shoot from mid-courtyard or the outer works. However, I'm still not completely sold on vision tracking for shooting. An inconsistent mechanism, or bad code, could make a vision tracking shooter useless in the blink of an eye. Shooting from midfield and the outer works presents the problem that you're not just fighting against defenders, you're fighting against yourself to have a shooter that lines up to the tiny goal perfectly and shoots at the right velocity, every single time. However, 179 and 2393 showed that it is possible to overcome these difficulties on a competition field, which is very impressive.

[BrendanB]

After witnessing the official Week Zero event and watching Palmetto I am surprised by how often breaching occurs. I think that is more attributed to most teams focusing their efforts on breaching (which is good) but that means teams had to focus less or not at all on Scaling or Boulders.

I think overall having to compete this close to bag day hurt a lot of teams but more importantly the first events of the year are always drastically different compared to Week 3 or 4 (I've mostly competed at Week 1 events in FRC until 2014).

I was very surprised by how many teams were successful in the high goal compared to the low goal. Overall I think that was mostly impacted by the low goal robots focusing primarily on breaching consistently (like 2200) instead of scoring and that the high goal robots made it their top priority avoiding things like scaling or damaging numerous defenses. 179 and 4451 were two of the most well rounded robots at the event tackling the high goal and defenses with ease but they are also some of the stronger robots in their region.

[team4243]

For us, visibility was a bigger issue than anticipated (drive team speaking here). If the opposing alliance selects both a portcullis and a drawbridge, teams without good cameras will have a difficult time. We saw one team do the pole camera for the standard, but we have no idea if that was helpful at all.

Breaching was crucially important, and two teams on an alliance should be able to do it every single time.

The FMS was buggy, (to be expected for a 0.5 competition) and we were totally incapacitated for one match through no fault of our own. Several teams, including 343 and others, had issues with communications dropping out during the match as well.

Unless your bot has an excellent shooter, it's not wise to spend much time lining up for a shot. As other commentators have noted, the low goal shooting is much more difficult than it appears, especially without a reliable camera. We had little success with shooting, and focused mostly on breaching defenses.

Although teams should already know this, the actual portcullis game piece is much easier to open than any wood imitations built for practice. It's on a coil spring, and one team had good success just by jamming a wedge underneath the door and driving through.

If teams within the alliance can cooperate, opening the doors is relatively easy, having one team open the door from the opponent's courtyard and getting the other alliance members to drive through from the neutral zone. We did this twice and it worked well!

Climbing will become more and more important as time goes on, and it was a huge factor in alliance selection.

Field reset times are much longer than usual, and the competition won't follow the schedule at all. Palmetto was about an hour behind during the entire competition.

Even if your team doesn't make it into Elimination, we recommend staying to watch. At higher levels of play, Stronghold is immensely fun to watch.

Stronghold is an excellent game, and teams everywhere should have a blast playing it. We wish everyone the best of luck!

Team 4243

[The other Gabe]

The major thing I noticed was that shooting into the tower was (mostly) useless in quals matches - teams rarely had enough consistent shooters to make that a better use of their time than breaching, which could both get you enough points to usually win matches, and another RP on top of that.

Then you get to Eliminations and Shooting is absolutely vital to your alliance's success, because everyone can breach now.

also huge props to 1369 for their amazing defense. It was so much fun to watch them completely shut down opposing alliance's shooting

[GeeTwo]

Originally Posted by PayneTrain

-Teams that designed for the high goal may have designed their robot to not low goal after it's brought into their high goal scoring mechanism. I wouldn't let Jesus take the wheel on that one, but teams decided they could live with it.

Yes, we designed such that once the boulder goes in the high launcher, that's the only controlled way the robot can get rid of it. However, our planned M.O. is to keep the ball in the loader until we're nearly ready to shoot high. There's a place in the intake where the boulder can be held securely, and we are using a range finder to identify that the boulder is in that spot and stop the intake. The intake holds the boulder more securely than the launcher, as well.