Is specialization or flexibility more important in robot design?

A member of my team and I got into an argument on whether it was better to attempt to create a robot that is capable of doing everything or make a robot that is specialized in a specific aspect of the game. When I say better, I mean a robot that performs more consistently at regionals/districts and nationals. I would like to get some second opinions on the matter.

Thanks in advance.

Here you go.

You cannot answer that question in a vacuum. It would depend how difficult your specialty was, the point value. and the rarity of robots be able to accomplish the task. The ability to rain balls in autonomous is highly valued in this years game. Just being able to climb not so much.

In general I would say robots that have the ability to do multiple tasks do better than one trick ponies. If you are going the specialty route,you need to the better than everybody else at the task and the task has a high value to toe game.

But on the other hand, if ALL your robot can do is rain balls into the high goal, you’re probably not winning a lot of quals, since it’s hard to make up the 90+ point deficit in gears and climbing. So yeah, it’s always a complicated decision, and a lot has to do with your honest assessment of your team’s capabilities. Flexibility is good this year, like every year, but a robot that does everything at a poor to middling level is going to lose to a good gear+climb bot.

I would say that a better approach at deciding which is better is to build a robot that you know will have the most positive impact in the match. For the best teams, this usually means a robot that can do any task since most top teams can effectively design and manufacture a robot that can do it all. In other words, they won’t be overstretching themselves with that choice.

For closer to average teams, as others have said, it really varies. It’s up to you to decide what the limits of your engineering prowess are. Sometimes giving up an ability will give you enough driver practice and testing time that you will play the game better with the more simple robot.

Just know, if you decide to build a specializing robot, you have to be prepared to execute a specific strategy throughout the competitions. You don’t have many options to vary strategy compared to a robot with more abilities. However, if you can build a strategy that complements other teams well, you will most certainly find yourself in eliminations no matter how you seed.

As for my personal recommendation, all I can tell you is erring on the side of simplicity is less likely to fail than erring on the side of complexity (where you can’t actually play the game well because the robot has too many issues or you didn’t get enough driver practice).

It’s best to make a robot that is within your teams capability. I would much rather have a good simple specialized robot on my alliance than a mediocre generalist.

Sent from my XT1254 using Tapatalk

If I’m a top team making my 2nd and 3rd picks, I’m going to be looking at filling a gap or augmenting a capability.

I can use my own team as an example. In 2004, we ignored pretty much all of the game (balls, goals, doubling balls) to focus exclusively on the end game. We designed our robot to focus only on the chin-up bar, and to play defense on it.

You can see us on Einstein below rolling along the bar preventing 67 from hanging. The final score of that match was 150-110, the defense on the bar helped to win the match.

This year, if you could chose a robot that could shoot fuel and score gears, but not pickup either versus a robot that could not shoot fuel but could pickup gears, (all other things being equal) what would you pick?
This year’s game was perfect to specialize, unfortunately too many teams (my own included) didn’t.

I agree to a large extent. It is hard to site an example because the robots that can shoot balls effectively are good in other areas. This shows the complication of this years game. It is hard to make up the 50 points of not being able to climb.

I know a ton of teams that built shooters and then haven’t used them once, because they realized they needed to focus on gears to win and contribute. Similarly, last year a ton of robots built high goal shooters and then ended up playing the low goal game the whole season.

Most of these teams would have been better off never pursuing high shooting at all and instead just getting a ton of driver practice and autonomous modes written. When the first overall selection for the South World Champions didn’t manipulate fuel at all this season, that’s how I know specialization was good for this game. And I’d argue that it’s good for almost every game, because the GDC tries to make to so teams can contribute to the game at every level (the most recent exception being Recycle Rush).

This year was the Year of the Specialized Mid-Tier Robot. Teams with a passive gear mech and a good climber did amazingly well- 115 made it to finalists at their division, and 604 made it onto Einstein (the Silicon Valley mid-tier in general was super strong this year). The lack of points from balls greatly incentivized making a good gear/climber robot, and if you specialized in that with a good driver you were a guaranteed pick/winner.
That’s still true for most recent years IMO. It’s harder to do that for years like 2013 or 2012 where there’s only 1 game piece/scoring option, for for years like 2016 and 2017 it’s possible to play the game without shooting and still do well.

At kick off we analyze the game. Before we start discussing the robot we consider points per second and what has the most value as an alliance partner. That might even change as the season progresses.

This year we didn’t specialize. Ran gears. played a little defense, climbed consistently, and by our last seeding match could shoot high goal for 3ish points. In terms of effort per point we spent a lot of time on the shooter. In actual time it was more balanced. In the last seeding game our shooting tied the match (not to take away all the other points scored). Our qualification alliance captain didn’t pick us for our shooting ability. (Thank for picking us Orbit). They had us tape over our hopper so we wouldn’t get a gear in it. :]

I’d argue it was easy to specialize in 2013 as long as you didn’t specialize in climbing. Shooting was pretty easy in 2013, especially if you gave up floor pickup, going under the pyramid, and climbing past 10 points.

I believe that there is a mid-to-top tier solution to every game in the modern era that requires a drivetrain, and a maximum of two moving parts.
2012 you could be a tiny robot with a bridge manipulator and a collector (4334).
2013 you could be a gravity fed high goal shooter with a 10 point climb.
2014 you could be a collector.
2015 you could be an elevator with a ramp/stacking platform (this one was really hard to come up with and execute, so it’s the weakest example here. See 1325.)
2016 you could be a collector with pneumatic wheels.
2017 you could be a passive gear mechanism and a climber.

It’s hard to remember that you can succeed at the highest levels with something really simple at the start of build season, especially when you think of all the cool possibilities.

Here is something that I have learned over the years that has helped my team (not necessarily anyone else’s just because of our team dynamic)

We usually chose 2 major features, I will use our last years bot as an example.

Last year our 2 major features were

  1. A drivetrain that could cross any terrain obstical
  2. a shooter that actuated such that if you reverse the shooting wheels and lowered the shooter it acted as an intake, and could shoot in high and low goals

Now there were several things this did not allow us to do in the beginning
cheval de frise (or however you spell it)
sally port
However what we do from there is modify what we had to be able to do some of those.

  1. we added aluminum flatstock to the bottom of our shooter such that it could go below the portcullis, and when the shooter would actuate into our robot to get into its shooting position it would lift the portcullis
  2. we also modified the bottom of the shooter to have 2, 1 inch blocks at the bottom to push down the cheval so we could cross
  3. we also relised that if we come from the back of the drawbridge with our shooter in the intake position, then have it such that only the shooter was touching the bridge, if we actuated the shooter then put it down quickly we could cross the drawbridge
  4. finally we also relised that if we just spin in the sally port we could do that as well

This leaves just one last thing we could not do, climbing. we were working on a climber after our second district event but did not have the time to complete it. Over the years I have relised this is kind of the limit to our team, small incremental changes can make a lot of a difference if implemented properly. nearly all of what I described above was done in between matches at our first district event last year so its really about knowing what your team is capable of.

blue alliance has us using this bot if any of my explanations were confusing. we ended up winning 2 districts and being and #3 alliance captains (and sub-division finalists) on our field at worlds, so the bot did well. I often find teams have trouble making these incremental changes due to having an incredibly complex bot, but in my opinion (unless your 254, 33, 67, 1114 etc.) it is often better to go back to the basics. my team for example is terrified of using pneumatic, so we don’t, and honestly that’s fine, as long as you can make up for it elsewhere.

here are a few matches if you are curious, or don’t understand the descriptions I provided earlier.

I really like this way of thinking about the game. Pick two major things, and excel at those. It kind of meshes with my examples of robots that succeeded at high levels with simple robots.
And clearly, it worked for you this year as well!

For 2013 that’s very true, but you would still need to have an accurate shot. 115 actually did really well that year (8th alliance captain at SVR) using an HP station loader and a good driver, but many teams couldn’t pull off shooting.
2014 you would be aiming to be a 3rd bot at that point. There were too many ok-good robots that year to be just an intake. Same thing for 2012; Even 4334 won as a 3rd bot (despite being 16th seed somehow). I would argue a step shooter may have been more competitive for that year.

I wouldn’t call it a somehow; they also ranked 13th at GTRE (their only regional) with #8 OPR, surpassed only by the likes of 2056, 1114, 2852, 610, 188, 907, and 1241. They were 1st pick of the 6th until 610/188/3360 knocked them out in the semis. Doing that plus ranking 16th and winning Archimedes in 7 matches has got to land somewhere in “mid-to-top tier”, right?

No; reliability is more important.

Lots of really good posts in this thread. The two cents I’d throw in is that any time you have a high-value task that will definitely require multiple robots to accomplish (like getting four rotors), you can pretty safely specialize in that task. Whenever there’s a limit to how many robots can specialize in that thing (like only having two HP stations in 2015), that’s a sign that flexibility could be helpful.

Just watching a few of their matches, it doesn’t look like they did anything particularly unique that can’t be accomplished by a mid-tier robot today, or by a high-goal robot from 2012; 3-4 low goals in auton and some ball transport, plus a tiny robot that could balance easily. OPR might be decieving at this point because they did not contribute huge amounts of points on their own, although paired witha good shooter they could do really well.
Maybe that was all it took to be good in 2012, and if so, hats off to 4334 for playing that strategy. I’m a personal fan of 701’s robot with the low shooter and the stacking robot strategy.

I would go even further: Pick two major things, provided that one of them is driving of some sort. (Look, driving is really important. Don’t build a drivetrain that has random failures. OK?) This gives you a drivetrain +1 specialization. If you’re really good at that specialty, you will do well.

Plan to add one other major thing after your first major thing is taken care of. So, build not at the same time but in sequence–but plan an attach point or three for your second specialization. If you build the interface in, you can have your second specialization during the time the robot’s in the bag…

If you still have time, you can maybe add in one minor thing. IF…

Basically, a specialized robot that adds more specializations is the way I’d go.