[FiM] Midseason Review

[FiM Informer]

Let me start by saying I have absolutely no intention of mentioning every good Michigan team in this post. We have over 400 teams, and I’m just a single human observer who competes with their team. I haven’t watched every match from every district, in part because there isn’t video from every district, but also because that’s impossible. I’m also not willing to post based on things I haven’t seen myself (event wins, OPR, whatever else). I will, however, try to mention teams I’ve noticed that might not catch your eye if you weren’t at their event. I’ll do that while hitting on a few topics that should be interesting/helpful for the world’s teams.

Playing early is an advantage

I feel like I harp on this every year, but playing your second event early gives you a big advantage, especially in seeding because even if you’re not great, you’ll dominate your opponents in early qualification rounds because many of them are moving their robots for the first time. Teams that have already played their second district averaged about 10 more points at the second one, which is a big chunk of points (equivalent to advancing to the next round of elims or a pair of regular judged awards) that you essentially get for free. Two Michigan teams have taken distinct advantage: 1701 and 2619.

At Southfield, 1701 was selected by the 8th alliance and proceeded to get beaten in two by the event’s top alliance (one of two capable of occasionally getting captures). The following week, the RoboCubs headed down to Indiana’s Tippecanoe District, where they led the event’s 3rd alliance to the finals. Coming back to Michigan in week 3 for the Center Line District, they were able to win the event (their first event win ever) from the #4 captain spot. At each event, they were able to make big improvements, taking advantage of driving that could be best described as enthusiastic as they transitioned from a primarily breaching role at Southfield to being a core shooter for their Center Line alliance.

2619 made similar strides in their path this season. At Standish, 2619 had serious struggles getting their intake arm to work consistently. They ended up with a quarterfinal exit on the 6th alliance as the 3rd bot. Even by elims they had improved significantly, and their alliance nearly took the series (one of the matches went to the 5th order match tiebreaker, which I’ve never seen before). Still, they had trouble with the portcullis and grabbing balls. These problems were gone by their Week 3 event at Midland, where they captained the 3rd alliance to victory and won the event for the second straight year. I also have to shoutout their qual 79 partner 5712 for making one of the more intelligent plays of the year pushing a dead partner onto the batter in the final seconds of a match to secure the capture and send both teams multiple spots up the rankings. 2619 could’ve played smarter at times; they’ve occasionally shown what I’d call coaching mistakes, like going to cross a defense that’s already been damaged before realizing, having wasted 5-10 seconds.

This is another sidebar, but Midland was a surprisingly strong event. Unless TBA is lying to me, they paced all Michigan districts in boulders per match and damage rates of 7 of the 9 defenses for qualification matches. (For the curious/lazy, aggregated TBA Insights data on Michigan events). At first, I thought this was because the event had 26 2nd event teams, but there wasn’t much correlation between those factors. Consider the other Week 3 events: Escanaba and Center Line had 9 and 18 2nd event teams, respectively, but nearly identical qual breach rates.

How do I win playoff matches

Basic Points

I’ll get into the points analysis of what it takes to win, but I feel the need to mention the obvious points first. Trigger warning for 1023 fans, but the first one is to not break. This game is pretty rough, and teams have been breaking down constantly. Beat up your practise robot, prepare for even the most catastrophic failure, and try to pick robust and reliable teams to join your alliance.

The second point is that everybody has to do their job, and everybody has to be ready to do somebody else’s job. While most teams aren’t directly interacting, they do interact through how well they do in their actions. What do I mean? In 2013, typical teams played their best ignoring everything but their own cycles. This year, you have to make sure that if your partner struggles with their defenses, that you make sure they get done.

A great example knocked out one of the better Michigan teams this year in 4967. They have a slick bot with a 20-pt auton but they lost in the QFs at St. Joseph. Looking at their first QF match, you can see in the breakdown they’d have won if they’d gotten two more crossings for the breach. The video doesn’t elucidate why that wasn’t achieved. At 80 seconds left, they’re 3 crossings away from the breach. 4967 headed to the sallyport around 15 seconds left, but that was much too late. Around 50-60 seconds, they have to realise the crossing is all they need to win and prioritise that, whereas 4967 is still scoring low and 2474 is still scoring high*. It was almost reminiscent of the final match of Archimedes 2012 (although at a totally different level), when the red alliance doesn’t realise all they need to win is a double balance and one scored ball, and instead they go for the triple and lose their shot at Einstein.

*I don’t mean to call anybody out. Mostly likely it was one team’s job and they forgot, or there was a misunderstanding as to who was supposed to do it. Everybody makes these same mistakes from time to time. Coaches this year need to be omniscient, seeing everything that happens on the field and adjusting gameplans as they go. Falling short of that is nothing to be embarrassed about, but it could cost you!

Captures and other Difference-Makers

The capture has been the focus of elims at many events around the world, but at several Michigan districts, the winning alliances haven’t had to rely on capturing to move on. Looking at the number of elim matches typically played and how consistently alliances are able to capture, a single alliance that can somewhat reliably capture contributes about 10% to the elim capture rate. Looking at the aggregate data linked above, that means capturing was needed to win 4 of the 10 Michigan events so far (including one of three Week 3 events).

Certainly captures will be more and more necessary as we move on. At all events, elim capture rate has gone from 13% (~1 consistent alliance) to 17% to 21% (~2 somewhat consistent alliances or ~3 who are on/off) these past three weeks. However, I’d guess there are still one or two of the remaining 11 Michigan districts where capturing really isn’t needed to win.

You want to maximise your chance of winning whatever the quality of the event. Let’s take the three possible cases here: (a) captures can’t really happen in elims, (b) captures happen all the time in elims, and (c) captures occur but they’re not a given.

In the first case, low goals become worthless. You’re not gonna capture? Accept it, and don’t shoot for it. Realistically, every (good) alliance is going to breach, and once you account for auton and challenging, that’s about 75 points that are a complete given. In this space, scores are suddenly real close. Small differences can and will make an impact.

Here are the difference-makers for your alliance:

• 3rd robot auton crossing
  One of the few difference-makers coming from your 3rd bot. These 8 points are worth a little gyro and code cheesecake. Although it should be noted that even this is becoming more of a given each week (most of Waterford 3rd bots had this capability).
• Getting the 5th defense
  Across the world, people are ignoring the sallyport/drawbridge, but they’re actually making the difference in matches without much boulder scoring. The sallyport is also pretty easy to do for any robot capable of spinning in place. For that reason, I’d suggest scrapping the sallyport in elims, even if the drawbridge obstructs your vision.
• Accurate (if low volume) high goal shooting
  We’ve been trained by FRC experts and MCC designers that the low goal is just as good as the high goal. When captures are out of the picture, that simply isn’t true. If you can shoot high accurately, you can double your cycle time and still come out on top.
• Scoring an auton boulder
  This here is one of the toughest items on the list, but it’ll make you stand out the most. Last year, you could almost guess before seeing any robots which teams would have a 3-tote auton. This year, lots of lesser-known teams have it (5448, 1701, etc.) while the two Michigan bots that attract the most attention don’t have it (33 and 67).
• Scaling
  This item is another attention-grabber, and it’s also maybe the toughest to add. However, this year teams that went low have created lighter robots in general, and that takes down one of the biggest barriers to adding a complex item like a scaler. Expect more of them to pop up, especially going into MSC.
• Defense
  I wasn't sure whether I should even put this here, because it’s mostly a given. However, good defense can stop opposing goal-scorers in their tracks (watch 5222 in elims against 5114 at Midland). In rare cases, a team could show off this ability in qualifications. Alternatively, you can watch for smart driving (e.g. maneuvering in traffic jams) and positional awareness (e.g. come back over defenses easily even without great visibility) to see who would make a good defender.

In the second case, captures are essentially a given for you and the alliances you’re competing with. I could see this happening at a couple events before MSC (possibly at Kentwood, Howell, Livonia, or West Michigan), but in absence of a 2771 that can put in most of a breach on their own, capturing is going to stay in the “it’s kinda hard” range. But to stay on topic, let’s say you’re guaranteed a capture. Suddenly, you’re looking at a list of difference-makers that’s exactly the same as the above. High goal shooting and scaling in particular are going to dominate at MSC.

The final case is that a capture is in the balance every match. In these cases, it’s tempting to tell you to sell out for the capture. That’s what has worked so far, and frankly it’ll keep working. But to this rule there can be exceptions. I’m reminded of a 2013 MSC alliance that really didn’t score many discs, but advanced to the semifinals based on climbing ability.

The point potential is there for an alliance to break the mold. Let’s say your opponents capture with 8 low goals. You’re not going that path, but you want to win anyway. You get 4 high goals instead, which puts you at net negative of 21 points. That’s a scale (+10), getting the 5th defense (+10), and a low goal (+2) away from a tie. For alliances at events with poor low goal scoring, this is a path you could go.

All that said, it’s hard and you’d have to walk into it in alliance selections (how many scaling high goal scorers are there that can’t low goal very well? Best example I can think of would be 280, who don’t even high goal particularly well). Additionally, there’s probably a fair supply of low goal scoring at your event. Your best bet is to treat the elim capture like you did the qual breach: your first task is to complete it and everything else is cherry on top.

Range classes for goal scorers

Right now there are four main classes of goal scorers, each with teams that are better and worse. Some teams are hybrids and can shoot from different locations, but even they will want to stick to one or two (they mostly have so far). So far, the low goal scorers (class #1) have dominated, with teams like 3602, 1481, 6193, and 6086 (who have a pretty inspiring story if you're interested) doing well. It’s telling that this is where rookie teams have found success. Being a great low goal scorer is more about simplicity and practise than having a technically impressive robot.

Teams are focused on putting lots of boulders into the goals, so efficient high scoring provides big point bonuses. As a result, low goal scoring will eventually go out of style. It’s hard to tell exactly when that will occur. My best guess is that it’ll happen right when you start playing against 3620, 1918, 67, or 33. These teams (class #2) shoot from the batter and are crazy consistent. This style of shooting could dominate up until and through Einstein. It’s possible to stop a team from getting to the batter, but you can’t stop two (unless you’re 294 from 2010 and can block two goals at once). Effective battering alliances could be those that (accidentally or intentionally) time their cycles and score in parallel to pull the defender’s attention in two directions.

The other classes are teams shooting from the outer works - essentially non-existent (I think I’ve seen 85 and 4967 shoot from the outer works? Like once each?) - and teams shooting from somewhere in between. The midrange shooters, including 2771, 2834, 3452, and 1701, will have a tough time unless they line up really quickly. They’re the most exposed to defense and have the least to line up against. I could imagine teams having success here, but it’ll be quite tough. Most of these teams are better off adjusting their shot (some can already) and moving elsewhere.

This post has already gone on longer than I'd hoped, so I'll let it trail off here. I may pop in to give a prediction of the points needed to make the State Championship. If not, I'll be back in three weeks for the MSC Preview. Either way, good luck teams as you go into the second half of the FiM season!

[ahaltom]

First, thanks for the shout out for qualification match 79 in Midland. As I am the drive team coach it was awesome to experience this match on the field.

After competing in 2 events, it seems the ideal alliance would be:

1. Consistent high goal shooter capable of 4+ boulders per match.
2. A robot that can pump many boulders in the low or high goal to guarantee a capture.
3. DEFENSE, DEFENSE, DEFENSE.

The winning alliance at Midland is a prime example of the combination.

1. 3538 - RoboJackets. Consistent high goal shooter.
2. 2619 - The Charge. Put many boulders into the low goal consistently.
3. 5222 - Jackbotics. Completely shut us down when we faced this alliance and shut down any other robot that was a high goal threat. Our alliance had no defense threat, so 3538 and 2619 were able to score boulder after boulder.

A defense breaching specialist may be nice to have as well, but from the two events we have been to, it seems like the breach just occurs naturally if you have capable robots. Also, if you are capable of scaling you will be picked no matter what rank you are after qualification matches (Example: 5166 at Midland).

Thanks for writing this analysis. I hope to see one for the MSC event!

[JABot67]

Great analysis, FiM Informer!

Do you think the batter shooters from MI have a chance to win against the 1678s and 2056es of the world who have dialed in shots from the outer works? What would be some key plays that would tip the balance to either side in that matchup?

Also, do you think it's possible for 67 to start shooting from the outer works? Their vision targeting has been improving, and I know they posted about redesigning their shooter to get a more accurate shot.

[tr6scott]

Good Read, thanks for putting that together.

[sodizzle]

UGH I'm so triggered right now! We didn't mean to break!!!! Haha just kidding
We've had two completely separate gearbox problems that I've never seen before, and from who I've talked to, they've never seen or heard of either. Hopefully we've fixed the problem (again) for this weekends competition.


But for real, great read. I enjoyed it. Thank you for the post.

[tr6scott]

^ I used to wish for TORC to be mentioned by FIMinfromer...
I'm ok with no mention this time. Be Careful what you wish for.

I have a feeling both TORC and BX will be back on tracks this weekend. Good Luck!

[Zebra_Fact_Man]

Originally Posted by sodizzle

UGH I'm so triggered right now! We didn't mean to break!!!! Haha just kidding
We've had two completely separate gearbox problems that I've never seen before, and from who I've talked to, they've never seen or heard of either. Hopefully we've fixed the problem (again) for this weekends competition.


But for real, great read. I enjoyed it. Thank you for the post.

If I may ask, what were your gearbox problems?
We had a gearbox problem of our own flair up the day before our first event, but it was diagnosed and fixed the following day.

[sodizzle]

Originally Posted by Zebra_Fact_Man

If I may ask, what were your gearbox problems?

At Waterford we were using the WCP DS Gearbox. We ended up destroying a dog gear, an axle, a shaft collar, and the mounting block.

At Pittsburgh we decided to get rid of that problem and used WCP SS gearboxes. Part of the problem was with the 11t pinions we were using not quite meshing with the other gears. I wasn't there so I can't say exactly what else happened but I believe the CIMS broke through the mounting hole that has 2 positions for the different sized pinions and spread away from the gear.

All that aside, we're ready to come out to Livonia with fixes and improvements and hopefully perform better than ever.

Originally Posted by tr6scott

I have a feeling both TORC and BX will be back on tracks this weekend. Good Luck!

I sure hope so! Good luck to you too!

[Richard Wallace]

Originally Posted by FiM Informer

..., low goal scoring will eventually go out of style. It’s hard to tell exactly when that will occur. My best guess is that it’ll happen right when you start playing against 3620, 1918, 67, or 33. These teams (class #2) shoot from the batter and are crazy consistent. This style of shooting could dominate up until and through Einstein.

Time to add 27 to this list.

[FiM Informer]

I was beaten to the punch on making point predictions, so I'm happy to defer to Eric on this one.

I'll be polling Michigan's experts again this year for the pre-MSC power rankings. Are you interested in participating? Fill out this application and I'll select a few guest voters.

Originally Posted by sodizzle

UGH I'm so triggered right now! We didn't mean to break!!!! Haha just kidding
We've had two completely separate gearbox problems that I've never seen before, and from who I've talked to, they've never seen or heard of either. Hopefully we've fixed the problem (again) for this weekends competition.

I needle because I care Was great to see BX back on track at Livonia.

Originally Posted by Richard Wallace

Time to add 27 to this list.

Most definitely!

Originally Posted by JABot67

Do you think the batter shooters from MI have a chance to win against the 1678s and 2056es of the world who have dialed in shots from the outer works? What would be some key plays that would tip the balance to either side in that matchup?

Also, do you think it's possible for 67 to start shooting from the outer works? Their vision targeting has been improving, and I know they posted about redesigning their shooter to get a more accurate shot.

There have only been a handful of events with world-class matchups, so it's hard to say which style of shooting will be dominant.

Let's look at batter shooting. It's easy to say that predictable shooting location leads to easy defense. Einstein 2014 is an interesting case study. A couple big factors made shots easier: open field, wide goal. Of course, a shooter had a minimum of one robot playing defense and often two. In most situations, a shooter facing a single defender (or two defenders but having a blocker) was able to break free of the defense and score, either by taking a quick shot (like 1114) or getting to the hard to defend place (like 254). The batter is a similar hard to defend place that makes lining up easier (especially once you consider the plastic barriers). Although the field is much less open, I think this year will be pretty similar in that even batter shooters should be able to get around defense with a small amount of difficulty, and should be able to get an accurate shot off once they're there.

Next, let's look at outerworks shooting. You mention 2056 and 1678 specifically, so I'll focus on them (folks who've watched them more than I have, feel free to chime in). They're both pretty d*** accurate from the outer works. I'd say 2056 is a whisker more accurate. The important point is that neither were perfect. 1678 in particular had a match at Sacramento where their shooting accuracy looking positively human. To me it seemed like their auto-targeting wasn't working. I'm deathly afraid of that happening at any particular moment, but especially on Einstein where conditions are very different from other fields. I imagine teams will get more time than usual on the field before Einstein matches begin, plus I'd bet those elite teams will have SOPs for retuning vision.

2056 is definitely great at the outer works shot, but when left undefended, they take a midrange shot, and I think that's telling. Any good team will tell you to minimize the number of places you take shots from. Why would they choose to take the shorter shot? Maybe it's a bit faster to line up, maybe it's really the same shot just from slightly closer, maybe it's significantly accurate. No matter what, it's an indication that outer works shooting is not optimal, it's just what they do to avoid defense.

It should also be noted that once your alliance has one batter shooter, the second batter shooter gets defended a whole lot less. Additionally, most of the outer works shots have been from directly in front of the middle goal (though with vision, it's not like they have to). If you have two teams that prefer the same spot, you could have some interference problems.

Altogether, I think batter shooting will be a legit strategy seasonlong. I'm not sure I saw the accuracy concerns 67 mentioned, so trying to move the shot back might be the real focus. It's always hard to redesign a system midseason because you have all the surrounding systems (in this case, the intake and shooter) as design constraints. I think it's also clear that vision targeting is absolutely critical for outer works shooting. Maybe 67 will have that in place, but they're also a team that had trouble crossing in auton every match, so you can draw your own conclusions.

[aeastet]

If you want to see a really good top goal shooter you should look at 107. They were the top shooting machine at the West Michigan Districts. Here are some of their videos:

https://www.youtube.com/watch?v=pBoaBdaGCrY&index=10&list=PL6fCSvDccI_6LFz E4X4srtWTr-UZDPye3
https://www.youtube.com/watch?v=FVL-...srtWTr-UZDPye3
https://www.youtube.com/watch?v=daMk0Wyfamg&index=25&list=PL6fCSvDccI_6LFz E4X4srtWTr-UZDPye3
https://www.youtube.com/watch?v=5H1u...DPye3&index=30
https://www.youtube.com/watch?v=BbZtFHfPoto&index=92&list=PL6fCSvDccI_6LFz E4X4srtWTr-UZDPye3
https://www.youtube.com/watch?v=-K5pMnXzXTQ&index=93&list=PL6fCSvDccI_6LFzE4X4srtWT r-UZDPye3

[Arhowk]

No love for my sophomore boys?

https://m.youtube.com/watch?v=zlREbAH2mas

[Zebra_Fact_Man]

Originally Posted by sodizzle

At Waterford we were using the WCP DS Gearbox. We ended up destroying a dog gear, an axle, a shaft collar, and the mounting block.

At Pittsburgh we decided to get rid of that problem and used WCP SS gearboxes. Part of the problem was with the 11t pinions we were using not quite meshing with the other gears. I wasn't there so I can't say exactly what else happened but I believe the CIMS broke through the mounting hole that has 2 positions for the different sized pinions and spread away from the gear.

All that aside, we're ready to come out to Livonia with fixes and improvements and hopefully perform better than ever.

Holy carnage, Batman!
1076 also went with WCP DS Gearboxes, and also wrecked a gear (the 42T gear that teethed to the 11T pinions). Our problem was 1/4 CIM motors was mounted to the outside hole (the other 3 were mounted correctly in the inside holes), and the pinion slipped and shredded away the gear it interfaced with. Luckily we found out about this problem during driver practice the Wednesday BEFORE our first event and got it fixed Thursday at the event.

I'm curious, why are your gearboxes and versblocks not protected by your bellypan? Seems awfully risky to leave $800+ of parts exposed to the field elements. We put our bellypan on for the sole reason of protecting our gearboxes. Invested too much money in them; didn't want to risk breaking them. I could easily see something getting hit by the corner of the rampart or the rough terrain and fracturing.

In no way am I trying to tell you guys what to do (1023 is clearly still miles ahead of the pack), but it does seems unnecessarily precarious, especially considering the rough nature of this year's field.

[Mike Schreiber]

Originally Posted by JABot67

Great analysis, FiM Informer!

Do you think the batter shooters from MI have a chance to win against the 1678s and 2056es of the world who have dialed in shots from the outer works? What would be some key plays that would tip the balance to either side in that matchup?

Also, do you think it's possible for 67 to start shooting from the outer works? Their vision targeting has been improving, and I know they posted about redesigning their shooter to get a more accurate shot.

Batter shooting really depends on your ability to get to the batter - if it's easy then we will do that every time because we won't miss. It may even be faster than a low goal for us now.

While outer works shots are defense free, line up time makes them no more advantageous for reducing cycle time than the batter for most teams (there are several exceptions) and even then the risk of missing is much higher. We'll be mainly batter shooting at Howell although our camera aim does function - we'll see what happens at MSC.