The missing feature: A common thread


#1

Throughout the years of FRC that I’ve been involved, I have noticed that there is one defining feature that sets apart the robots of powerhouse teams and regular ones. Every year, there is some feature that the powerhouses realize that the rest of the teams miss. This year, it was over the bumper collector. Last year, it was roller claws. Breakaway was active ball control mechanisms. Towards the goal of furthering my team towards the powerhouse route, I’ve racked my brains trying to come up with an answer and a solution, and have come up with none. So I will phrase the question this way:

How do these teams come up with these ideas?

Why do other teams not come up with them?

The other question is, are these teams powerhouses because they come up with this kind of idea? Or do they come up with this kind of idea because they are powerhouses?


#2

For our team, we actually ruled out the extended drop down intake/bridge arm because we thought it would be best to simplify our rookie bot a bit, but it was in our initial design.

We later ended up adding these features anyway :rolleyes:

Overall I think it just comes from experience and a good design process. Also, collaboration with others helps A LOT.

Also a lot of the time there are normal teams who will come up with something like drop down intakes whathaveyou, but don’t execute them as well as the elites.

I think for this season, the real feature that put a couple teams at the top was a balancing aid + drop down intake.


#3

Yes.

:wink:


#4

How about the implementation of those ideas? Ideas are relatively easy compared to making them a fully-functional reality.

I think probably the best way to come up with ideas like that is prototyping and/or game simulation. Simulating the game tells you that X will be important; the prototypes tell you what highly effective methods you can come up with.

Sometimes, it’s just the execution, though. In 2005, the powerhouses tended to go for single-tetra robots, built for “surgery”–they could get in, cap a row in a particularly awkward place for an opponent, and get out to load up and strike again quickly and cleanly. Other robots built single-tetra robots, but they couldn’t pull that kind of maneuver cleanly and reliably. (The multi-tetra robots had a much harder time; I actually saw one single-tetra robot clear a stack of 3-4 off the top of a multi-tetra robot just by hitting the tetras.)


#5

I’m glad you posted this, because I’ve been thinking exactly the same thing for the last couple weeks. I haven’t been part of any top-tier teams, so take my thoughts with a grain of salt, but I think there’s several things at work:

  • Powerhouse teams have a very strong understanding of the playing field, because they take time to build a nearly complete one and prototype with it. I can’t tell you how many teams at the Oklahoma regional had bridge manipulators which were completely unable to push the bridge down, because they hadn’t actually tested on a real bridge.

  • Powerhouse teams are focused on performance. This seems obvious, but if you look at the top echelon of robots, they have one thing in common: they’re fast. Unbelievably, jaw-droppingly fast. Great teams design their robots to do the difficult tasks quickly, and squeeze the most performance possible out of their motors and mechanisms. When a team moves from “what’s the easy way to do this” to “what’s the fastest way to do this”, they move up in the standings.

  • Powerhouse teams try a lot of things. For a while my perception was that great teams were great because they had a group of geniuses who could CAD the perfect robot and a machine shop to fabricate it. But as I’ve watched and read (particularly 148 and 254’s build blogs), I realized there’s a lot of prototyping, testing, iterating, and tweaking that goes on.

Obviously, these teams are “powerhouses” because they come up with these defining features year after year. But they develop these features because they are relentlessly pursing high performance designs. They aren’t content with the first idea that comes to mind, or the idea that will be easiest to fabricate. Nor are they content with the first thing they build.

After watching a regional, most teams could go back home and design a really good robot, but by then it’s too late. The trick that powerhouse teams have mastered is doing this type of learning and iteration during the build season - before they get to the regional.


#6

A lot of it comes down to speed and iteration: most teams finish at least part of their robot a few days before ship, use it, and it breaks so they improve it. Then it’s time to crate or bag the robot, so that’s all the improvement you do.

Powerhouse teams got to that stage weeks before you did, and so have found problems or places to improve, improved them, and so end up with a much better robot overall.

2702 example: we had our robot completely done about a week before ship this year. In that week, we fixed a huge reliability problem with our ball-lift, discovered and fixed an encoder problem on the shooter, discovered the robot was too tall/tippy and shortened it, changed the drive code for further tip resistance, and redesigned the pickup software. The robot we bagged was at least 2x as good as it had been the week before. Now imagine 2-3 weeks of that kind of improvement, and you have a powerhouse robot.

Also they’ve got really smart and experienced mentors and students, so their first iteration may be better than yours. Plus they often have many sponsors, so they can iterate faster than you. But both of those things a smaller team can, over time, acquire.


#7

I’d like to delve a little more into the “fast iteration” thing. We also put a lot of time into prototyping, but never seem to do it as quickly or thoroughly as the powerhouses. What are the factors here? For example, do we need to find some kind of fast-fabrication sponsor? (I’m certainly not implying that’s all there is!)


#8

I think if you have the space for it, saving old mechanisms or pieces of old mechanisms can help with this. Without knowing anything about next year’s game, I can guarantee you that robots will need to:

A. Drive around; possibly over obstacles; possibly with pushing and shoving
B. Pick up and/or move around objects; possibly of odd shapes; possibly very heavy
C. Place and/or throw those same objects

In my discussions with people on powerhouse teams, they spend a lot of off-season time on “what if’s” – what if we have to shoot a ball? Lift a tube? A crate? A tetra? Another robot? Our robot? What if we’re playing with those pieces, but aren’t allowed to pick them up? What if we can’t break our own frame perimeter? What if we can?

Add in height limits, carrying limits, and anything else you can think of. Research every single FIRST game and come up with ideas on how you would create mechanisms using the resources you actually have that are fast, durable, and reliable.

The tasks that robots are required to do have not changed much over 21 years (move self; move various stuff). Coming into kickoff with a stable of ideas that can be iterated is much better than coming into kickoff with a set of skills but no solid ideas…

…and that means that when FIRST throws you for a loop (orbit balls, slick wheels, not being able to pick balls up off of the ground, absolute height limits), you have more time and brainpower you can direct that the novel tasks, because the less-novel tasks are well in hand.

(All that sounds good in theory. Somehow 1551 still manages to seriously screw up something on a yearly basis. Perhaps next year we won’t!)


#9

(1) I considered the drop down intake to be something that would only compliment an already great robot. I figured that if we could make that ‘great’ robot, then adding the drop down later wouldn’t be an issue.

(2) The team felt that the mechanism was also unnecessary. In years past where the bumper zone has been limited, teams have made do with the openings that have been available to them. This is where I think we struck out. Since I supported this from the beginning, I’m partially to blame.

(3) We simply do not, and would not, have the weight even if we wanted to. This comes with the territory of semi-questionable design practices, using reckless amounts of 8020, and simply not planning far enough ahead.

  • Sunny G.

#10

No one has said practice time?

As StevenB said, powerhouse teams build (or perhaps invest in) accurate fields. As Bongle said, powerhouse teams build fast. As pfreivald said, powerhouse teams design with years of experience (theirs or otherwise) behind them.

But the real key is being able to practice. Sometimes that’s just about finishing fast enough to have that week in the end of build season to drive around, or to go to a scrimmage. Sometimes that’s about building a practice bot. But practice helps teams work out kinks and discover problems they never anticipated or new uses for existing mechanisms. That’s why teams usually get better from one regional to the next.

Powerhouse teams start out with the benefit of a solid shakedown of the robot and are thus a leap ahead of the competition.


#11

This is definitely helpful. But I have to say, we have a well-ish equipped practice field and a practice robot, and we still kinda suck, especially compared to the perennial powerhouses. :o Yes, we get better, but no where near to their extent. And given that practicing with your robot is inherently a very late step in the process, I’m willing to bet we’re missing some of the early skills. Even if we’d put together that we should use a lower and better controlled shooter and an over-the-bumper collector when we started practicing, we still would have been SOL for a while. These guys figure it (“it”) out early and then practice and refine it even more.

This is a great discussion, by the way. I’m trying to put together a file of the top robots (photos) from prior years to see if there’s more to be learned on the “missing feature” front. I like the thought that a lot of it is about off-season “what if”-ing and keeping testbeds around–if only because we do that and theoretically will continue to improve at it as we pour or effort that direction. I hope.


#12

I’ve allways believed the key was building a robot not to play the game, but to play a strategy. Then you break down the strategy and figure out the quickest, most efficient way to execute it. The over the bumper intake was successfull because it allowed instant acquisition of balls and removed the variable of the carpet, as did roller claws last year.


#13

I think the over the bumper pickup was actually more successful because it allowed you to build a robot with a larger pickup, and therefor made it easier for drivers to pick up balls. We have a through the bumper pickup, and the ball made contact with the ground for the same amount of time as any over the bumper pickup.


#14

I agree, I thought he was comparing over the bumper to through bumper. I find it a bit interesting that so many elite teams went with over as opposed to through bumper.


#15

Experience is a great tool to have, but a better tool to have is funding and enough staff to implement it. If you have funding and staff you can by the extra material, build different prototypes and develop them into a perfect precision machine. With that funding you can build two robots and as one is bagged you can practice and develop the software and drivers ability. Having funding also can help if you can build a full field to do the practicing on. Once you have that you can make sure you pick your competitions so they are a few weeks in to let you work on the design and than your second competition is a few weeks later so you can fix from what you have learn at the last competition.
We have plenty of experience doing this for over 12 years and we know what we have to do but our funding limitations keep us from the best possible robot. Our work space has a low ceiling, we do not have a field to test on, we try to build some of the game pieces but it does not match a full built field. We have no professional engineers and our programmer has a full time third shift job. He has worked with students in programming but they are limited in what they can do.
We come really close this year to beating the top ranked teams and we are proud of our success. What we found is that the teams are labeled into two groups, the scoring group and defense group. If your robot does not do well you are termed a defense robot and pushed into playing defense during game by the experienced teams and than during the finals you are only picked if you are all that is left. We are experienced at this. A few years ago we decided not to be a defense robot and threw out low power and low speed. We design (to the best of our ability) or robot to be fast and focus on the highest points in the game. Last years game the highest points were the top row so we tried to design for that, this year the heist points were the top hoop and the bridge control. To get these points you must be fast and accurate. With the limited resources we could not build multiple shooters and found that because of the differences in the balls we could not shoot consistently to the top hoop. It would have been nice to find this out before we went to a competition but you deal with it the best you can. We did solve our problem but it was not nearly as perfect as the top teams. We will still keep working hard to beat the top ranked teams and be thorns in there play during competition.


#16

The first item is not necessarily true, while the second one is absolutely true. We build some of a wooden field, and even then, not until week 3 or 4. BUT we IMAGINE a perfect field and… (see below)

No, you just need to prototype faster. Kids tend to work slowly, we teach them how to work fast.

Yes indeed.

Also yes indeed.

The very first step in winning the game-of-the-year is to work out how the game will play. We ALL know how Baseball is played (for example) and so oddball strategies like a Bunt, which aren’t at all obvious, can win a game.

We take the rest of Kickoff Saturday and play the game many, many times, learning as we go. What do the robots need to be able to do? Anything subtle we need to think about? Then we send everyone home to think on Sunday.

Monday, we start by defining Capabilities. we determine which we need, and by Wednesday we can start on methods to achieve Capabilities. Then comes the prototypes - dozens of them. They are demonstrated by Friday (yes, FAST prototypes).

Meanwhile, the drivetrain team has a rolling platform ready for Saturday. The drivers drive, drive, and drive some more.

By week 5, the practice bot is driving and playing, we’re assembling the competition robot.

In the whole process, the simple things like scoring are designed to be simple and therefore quick. Any task that can be simplified, is. We want the drivers to focus on the things that can’t be predicted, and the mundane tasks are either automated or very simple.

This year, shooting and ball gathering is brainless, just drive to the right spot. Pushing down the bridge involves A) deploy mechanism and B) drive up to bridge. It lowers and you drive up without stopping. We go to ultra-low gear for balancing, but the balance itself is manual.

Kind of like modern aircraft cockpit design: Reduce the pilot’s workload so they can focus on what is critically important.


#17

No hints for the lesser OPRs? That’s ok, I’ll be happy to pick your kids brains at Philly. This is definitely the key I haven’t been able to make work. Good to know we don’t necessarily need off-site help. Thank you, Sir.


#18

In my experience, fast fabrication correlates to high performance. It doesn’t really matter if it’s through a sponsor or the team simply has a lot of machines at their disposal, but being able to go from the design phase very quickly to the assembly phase seems extremely important.

Our team has a manual lathe and a CNC mill in our machine shop which, is not really comparable to the resources at the disposal of “powerhouse” teams. In order to have enough time to manufacture a practice and competition robot, we had to finish the detailed CAD by the Monday of week 2. For us, time is simply that much of an issue.

Contrasting this with 254 (for me, the local powerhouse team) reveals a large difference. Talking to mentor, I learned that they typically finished the CAD around week 3 or 4. This gives them approximately two to three times more time to prototype, strategize and design. It’s hardly a surprise how dominant they were at SVR.

I don’t think that this trend of designing until late and manufacturing fast is unique to the poofs. To me, being able to get parts fast allows for extra time to prototype and design for the “powerhouse” teams that “normal” teams don’t have. Quick manufacturing turnaround, to me, makes all the difference.


#19

I also haven’t people mention communication.

I feel like the powerhouses talk to each other. I’m not saying there’s a magical forum where only the powerhouses post, but just through mentors knowing each other, students knowing each other, and alumni moving back and forth, the powerhouses must share ideas.

So at the end of the day, you’re not just seeing the best robot built by team xxxx, but you’re see the product of ideas from team xxxx, team yyyy, and team zzzz, but each team takes this feature and implements it differently.

This will obviously not play as much of a factor as having a full field or having quick machining capabilities, but I feel like somewhere down the line, talking out your ideas with multiple world class teams/mentors doesn’t hurt either.

  • Sunny G.

#20

Well that’s the interesting thing. These ideas don’t appear on Chief Delphi until week 2 or 3. Which is interesting.