The Quest for Einstein

[Cory]

Originally Posted by Basel A

16 made that alliance. 16 was one of the 10 best robots on Galileo, and they were the 21st robot into the elimination rounds. The fact that they were around is slightly more ridiculous than 973 in 2011 and slightly less than 1671 this year.

Doesn't change your point, but 973 was like the 4th best robot in our division in 2011. If we had been in a scenario where 1 forced 111 to decline and picked 1114 next, we likely would have taken them 2nd overall. Them being around was far more shocking than 16.

[Citrus Dad]

Originally Posted by tindleroot

This year should have been great, but then every team decided to make stacks by themselves IMO the best alliance possible would have a super tote stacker with a separate can topping robot (such as 148's two robots), but there were so few opportunities for this since most teams found that they can just build the whole stack underneath the cans.

I suspect that the GDC may have thought that building stacks would take more teamwork, and didn't anticipate the success of individual teams in solving the problem. That there was such a wide disparity in the ability to build multiple stacks may illustrate how technically difficult the problem was. The solution by the GDC might have been to give bonus capping points if a stacked was capped by a different team.

[Monochron]

Originally Posted by Rangel(kf7fdb)

Just depends on what you decide to specialize in and the execution. Some specializations also work better in the regional level vs championship level and vice versa.

It also depends a bit on what region you are in or how early in the season the competitions are. For instance at Palmetto there were elimination matches with only tote+can robots and tote robots. Cans were left on the floor while the teams racked up tote points but ignored the cans. There were at least 2 working can specialist robots that I remember who sat out of eliminations for whatever reason.
It seemed to us that can specialists were not highly valued, in part because there were so many tote+can robots to choose from.

[waialua359]

Originally Posted by Citrus Dad

I suspect that the GDC may have thought that building stacks would take more teamwork, and didn't anticipate the success of individual teams in solving the problem. That there was such a wide disparity in the ability to build multiple stacks may illustrate how technically difficult the problem was. The solution by the GDC might have been to give bonus capping points if a stacked was capped by a different team.

Interesting point.
Even after week 1 and 2 events, I would have still doubted that by season's end, the abundance of teams that could do 2 or more 6-stack capped noodled containers.
The game would have looked a whole lot different with your suggestion, something that forced everyone to take into consideration just one year prior.

[marshall]

Originally Posted by Monochron

It seemed to us that can specialists were not highly valued, in part because there were so many tote+can robots to choose from.

What do you mean we weren't highly valued? It's not like we had to completely rebuild the robot from the ground up or anything.... ohh, wait...

[nuclearnerd]

Originally Posted by Citrus Dad

I suspect that the GDC may have thought that building stacks would take more teamwork, and didn't anticipate the success of individual teams in solving the problem. That there was such a wide disparity in the ability to build multiple stacks may illustrate how technically difficult the problem was. The solution by the GDC might have been to give bonus capping points if a stacked was capped by a different team.

I really like this idea. Had that rule been in place, we would have designed our bot differently. We didn't even try to cap stacks (we can't even lift a can above more than 2 totes), because making our own capped stacks was the most reliable way to seed high. It would have been a lot more fun to watch teams cooperate on stacks rather than doing their own thing independently (the same every match).

Assist points really focussed alliance strategy in 2014. Hopefully the GDC brings back something similar for 2016!

[logank013]

Originally Posted by Citrus Dad

I suspect that the GDC may have thought that building stacks would take more teamwork, and didn't anticipate the success of individual teams in solving the problem. That there was such a wide disparity in the ability to build multiple stacks may illustrate how technically difficult the problem was. The solution by the GDC might have been to give bonus capping points if a stacked was capped by a different team.

That's actually a very good point. It would alter how alliance selections happened. It would have changed our whole robot design probably. We were one of the most successful short robots this year. We could only cap a 2 stack with a can. We finished around 40th opr wise and 3rd on Archimedes but most of the top robots above us were really tall. Besides 2826, I can't think of any other very successful short robots. (I may be wrong). But if that rule was added to the game, it would have made Recycle rush more interesting to watch, more suspenseful (having another team cap your stacks would be scary), and a whole new game to scout. Needless to say, it would probably mean a bigger range between the top top teams and the good teams.

[JesseK]

Our story this year echos a lot of what's been said in this thread. I don't know if there's a good football analogy for it though.

In between our last event and champs, we doubled down on our niche of can manipulation. We also tuned our Canburglars so they worked on-field rather than just in practice. They weren't the fastest in the world, but were definitely as fast as we could envision at the time given our robot's setup (not having seen 254's bolt-ons, that is). This played well with our claw that could up-right a can, as well as mods to ultra-grip the Can as we sped around the field to noodle it and set it exactly where it was needed. The decision was to only focus on those two things, ignoring all totes. It required re-framing our strategies to what a partner needed, rather than how we provided the most value carte-blanche. If we got an alliance which couldn't do totes - our job was to still ignore totes because we were a can specialist (though we could do a single coop tote with the claw).

We didn't even re-assemble our tote stacker at champs. For 4.5 weeks of effort, that was one of the toughest decisions our team has ever made. The strategy was simple: get 2 cans in autonomous, noodle them, place them, and stay out of the way unless something else was discussed.

As a mid-pack robot, it was easily the best decision we could have made going into a high-profile event. At the end of Friday, we had a good feeling that we were among the best in the division in something that the all-star robots could do, but would prefer to not have to waste the time on themselves. We also felt good about being one of the very few division robots which could reliably get both cans from the step each match all throughout the 2nd day (missing 1 can in 5 matches Friday, iirc). Slow (600ms touch, 2 seconds off the step) comparatively, but still effective.

The good indicators: we were repeatedly scouted by 3 of the top 8 teams towards the end of Friday, two of them with very specific questions.

This isn't the vision we had starting out on Day 0, kickoff day. For some on our team Wednesday at champs was a real struggle of anxiety. This type of thing isn't easy to envision, think through or execute. It requires new mechanisms, ways to field test them and a lot of preparation going into the event. Yet it's 100% completely worth it when it changes the champs experience for the team.

[apache8080]

Originally Posted by Karthik

Not that many people here will care about this line of discussion, but I think that the "Trent Dilfer" strategy might be more appropriate. Joe Montana may not have had the same gifts that Dan Marino or John Elway had, but he was still going to be a top tier quarterback wherever he played. (Take his end of career stop in Kansas City for example.) Trent Dilfer on the other hand was a very average Quarterback, but still won a Super Bowl while being an efficient "game manager" for the 2000 Baltimore Ravens, while relying upon strong running game (Jamal Lewis & Priest Holmes), and one of the greatest defenses of all time.

I think that comparing average robots that end up on historically good alliances to game manager quarterbacks is a bad comparison. Majority of the robots in the 2nd pick on top alliances usually do one specific thing on the field really well. I would compare this to being like Julian Edelman or Wes Welker in football or Matt Bonner in basketball. These are players who are overlooked because they don't have the obvious physical attributes but do small but very important things well like route running or shooting threes.

[sonichammer7476]

Originally Posted by BrendanB

I think the better goal here is how to move forward to become a consistent high performing robot at the Championship not just a goal of being on Einstein. Let's face it there are many, MANY amazing robots and teams each year who don't make it to Einstein because so many little elements determine where you seed, what alliance you end up on, and who you face.

This is very true. Our team is going on our 5th year, and have made it to Worlds for the past 2 years. Last year, we ended up being ranked 18th (or somewhere close to that) in Curie, and were not selected to advance into the final rounds. To me, we earned that ranking of 18th not through luck, but through a fair amount of skill (this greatly contrasts our first Worlds appearnace, where we were not blessed with a kind schedule). Many of the Top 10 selected teams with robots that they deemed 'expendable', in the sense that they had full intentions of butchering the robot to add elements useful to them, e.g. claws for cans. Sometimes it is not a concern of skill; rather, it is the fact that you are a 'tweener', like us (not Elite yet, but well on our way there).

[Citrus Dad]

Originally Posted by sonichammer7476

This is very true. Our team is going on our 5th year, and have made it to Worlds for the past 2 years. Last year, we ended up being ranked 18th (or somewhere close to that) in Curie, and were not selected to advance into the final rounds. To me, we earned that ranking of 18th not through luck, but through a fair amount of skill (this greatly contrasts our first Worlds appearnace, where we were not blessed with a kind schedule). Many of the Top 10 selected teams with robots that they deemed 'expendable', in the sense that they had full intentions of butchering the robot to add elements useful to them, e.g. claws for cans. Sometimes it is not a concern of skill; rather, it is the fact that you are a 'tweener', like us (not Elite yet, but well on our way there).

I point out that this year was very unusual in what mix of robots worked best on the field. The game was not well designed to use all 3 robots simultaneously in a high performing game. In fact, you could even see it in our alliance. Because we had 1671, we simply ramped back on how many stacks that our other two bots had to put up, down to 2 instead of the 3+ we had been making in the qualifying rounds. There weren't enough cans to make it worthwhile to go for more.

In previous years, the third bot could fit any number of roles. In our 2013 and 2014 alliances, our second choice could have been a first choice on lower alliances (and we were surprised that they were available at 24th pick.)

I'll point out another aspect that is key to the value of the third bot. We call it "value added"--its scoring non-teleop goals (which usually done by the first 2 bots) plus defensive ability. That's scoring in auto and the end game, plus stage points such as the assists in 2014. We always rank our 2nd picks by dropping teleop goal scoring.

The message is work to be in the top dozen and focus on performing supporting tasks that gain value to an alliance. Don't focus on trying to build for the "star attraction" of scoring the final points like goal scoring. Instead think of all of the ways to score points and think of which ways are least likely to interfere with the top alliance captains during teleop (or even auto).

[nuclearnerd]

Originally Posted by Citrus Dad

The message is work to be in the top dozen and focus on performing supporting tasks that gain value to an alliance. Don't focus on trying to build for the "star attraction" of scoring the final points like goal scoring. Instead think of all of the ways to score points and think of which ways are least likely to interfere with the top alliance captains during teleop (or even auto).

Good advice dor many, but Is that really helping the OP? They are asking what it takes to become an Einstein calibre team, and AFAIK, "Einstein calibre" teams get there reliably by building high-scoring robots, full stop. I guarantee 1678, 1114, 254, 148, 118 etc (to name a few) don't plan to fill a niche for someone else's alliance.

That said, it takes a lot of capacity building to get to the point where you're capable of consistently fielding robots that seed high. Unfortunately I can't offer any advice on how - we're not there yet

[Scott Kozutsky]

Originally Posted by nuclearnerd

I guarantee 1678, 1114, 254, 148, 118 etc (to name a few) don't plan to fill a niche for someone else's alliance.

I wouldn't be so sure. Just try saying that in the presence of tumbleweed, (arguably) 148s most iconic robot and the best example of a niche bot there is (that i'm aware of).

[Knufire]

Originally Posted by nuclearnerd

Good advice dor many, but Is that really helping the OP? They are asking what it takes to become an Einstein calibre team, and AFAIK, "Einstein calibre" teams get there reliably by building high-scoring robots, full stop. I guarantee 1678, 1114, 254, 148, 118 etc (to name a few) don't plan to fill a niche for someone else's alliance.

That said, it takes a lot of capacity building to get to the point where you're capable of consistently fielding robots that seed high. Unfortunately I can't offer any advice on how - we're not there yet

It's all about taking baby steps to get there. In my opinion, one of the biggest accelerators in developing on-field success is, well, on-field success. You could argue that 1678s big break was winning their division in 2013, and you can see how they've gotten much better since then.

Even if you realistically think you can compete at that level, games can still blindside you. Many, many top tier teams tried and failed to build a robot that could do everything in 2013, for example. You'd be surprised how many teams had higher hanging mechanisms that just didn't ever make it on the robot due to time or weight; it was a definite wake-up call.

On another topic, there's been a lot of talk about how, for teams that can't compete at the same level as perennial powerhouses, their best chance at going far at the championship is by building specialized support-role robots. I agree with this, but recognize that this comes with a very big assumption: the goal is to go as far as possible at the championship. Often, this robot won't be quite as successful at the regionals, depending how competitive your region is. If you were in an area where most people were only putting up a partial stack of totes, a can specialist robot wouldn't have nearly as much use as they would in California or Michigan where they would have much better robots to support.

[Citrus Dad]

Originally Posted by nuclearnerd

Good advice dor many, but Is that really helping the OP? They are asking what it takes to become an Einstein calibre team, and AFAIK, "Einstein calibre" teams get there reliably by building high-scoring robots, full stop. I guarantee 1678, 1114, 254, 148, 118 etc (to name a few) don't plan to fill a niche for someone else's alliance.

That said, it takes a lot of capacity building to get to the point where you're capable of consistently fielding robots that seed high. Unfortunately I can't offer any advice on how - we're not there yet

I think we've built niche robots in each year of 2013, 2014 and this year. In 2013, we needed a FCS to be fully effective (see 148). In 2014, we needed a goal finisher (see 1114). This year we focused on cangrabbing and were not extraordinary at stacking, just very reliable. None of our robots were extraordinary compared to, to name a few, 1717, 2056 (2013 in Curie), 1114, 971, 16 or 358, 254, 469 (2014 in Newton), 118, 1114, 254, 148, 987 (2015), and I can name several other robots that were of similar quality to ours in those years. We built our robot to fit a role in an alliance--just good enough. That's achievable by several other teams just in Northern California, and certainly other teams around the world. We started in our focus in 2012 and weren't of particular note before that. We were still working out of shipping containers in 2013.