Why so heavy, and who cares about free speed?: Poor design metrics in FRC

Technical Technical Discussion
1

Alright, my foray into FSAE has me really questioning a practice in the FRC world:

Building 120 pound robots rather than 60 pound ones.

It kinda hit me back when I was still in FRC and learned that 254’s 2014 bot weighed in at only 90 pounds, and then seeing that they were so effective on field in offense and defense not because of pushing power but because of speed and agility. Plenty of other examples of this are out there.

Additionally, COF of wheels/tread isn’t constant because tractive force comes from both adhesion (which doesn’t vary with normal force) and friction (which does)- a lighter bot will actually have a higher COF if I’m not mistaken. This is the case with road car tires, at least; maybe this is inverted with carpet (anyone done any testing?). If grip/pushing power is your concern, get testing different compounds and build tire warmers for your bot so you get a better COF.

So why build heavy? One justification is always ‘pushing power’, but I don’t buy that as shown above. If you can beat your opponent on speed, you don’t need to push them.

Recall newtons’ second law: F=ma. Remember this formula. It is the fundamental law for all that you do on the robot. If you want more acceleration- whether that be of your whole robot or an arm, you need either more force, or less mass to move.

6CIM and <big number>775Pro drivetrains have been shown to really drop battery voltage. Per motor, they’re doing less than 4CIM and <smaller number>775Pro drivetrains. Power (force) output isn’t limited by traction, but your battery- so if you want to accelerate faster, then you need to cut mass. F is at the limit. You’re gonna have to cut m.

‘But Thad, we need all these mechanisms on our robot!’ yes, you’re overbuilding them or building them with ■■■■-poor load paths, I’m sure of it. :]

There’s an old adage that ‘removing mass means you can remove more mass’ and it’s true- if your mechanism isn’t as heavy, it doesn’t need to support as much force from its own weight and acceleration. Thus, you can lighten it up more.

So when I saw gearbots this year being dominant, I was always astounded that they weighed anything more than 60lbs for a drivetrain with a ramp or snatchy arm. Stop hauling around the extra mass you’re not using and accelerate faster.

How, and why are these decisions being made?

I think part of this is teams using metrics like ‘drivetrain free speed’ or ‘drivetrain 85% adjusted speed’ to measure the performance of their 'bot. In reality, we rarely see these speeds on the field and spend the majority of the time accelerating up to position. A bot with lower top speed but higher acceleration can prove itself to actually be quicker on the field once defense is considered. Shifters are added with a ‘lower speed’, but we know why the shifter is added- it’s to increase acceleration and push the limits of grip for pushing matches! So why then, is speed the metric by which these systems are compared?

The other bit is the basic ‘how a motor behaves’ isn’t really thought about. There’s things like ‘you want to target 50% motor speed because maximum power’ or ‘you want to target 80% motor speed because maximum efficiency’ but that’s all steady-state operation- when the motor speed isn’t changing. There’s only two systems on FRC bots that operate in steady-state condition:

  • Air compressor
  • Flywheels

For everything else, the motor is accelerating and decelerating. You must look at a curve. VexPro even has those nicely for you.

Many just think ‘throw more power at it’ but forget about F=ma, voltage drop, and so forth. They could build a lighter mechanism with their motor operating with the correct gear ratio but choose not to. Winning FSAE cars make under 50HP; losing ones might make 90+ HP. Why do the winners do better, then? They weigh nothing and so have better grip.

Someone will tell me ‘these kids don’t know calculus’ but that’s no excuse for not being able to deal with physical reality because:

  1. There are plenty of tools on CD that deal with things like ‘sprint distance’- both numeric and analytic.
  2. Basic understanding of ‘the slower the motor goes, the more torque it makes’ would go miles towards understanding physical tests.
  3. There are kids who know calculus.

And yes, it’s more than just drivetrains. I saw a really cool example of a lifting mechanism by an FTC team (I want to say it was Got Robot) that used some cam-shaped spool because they understood the transient nature of their system: the motor as a result ran at max power (near steady-state) for most of the time and the lift was very fast. I don’t think they did any calculations but simply understanding the principle of how the motor worked helped them come up with a design that achieved the ultimate metric: time efficiency.

Overall, I’m just a bit frustrated to see entire groups of people in ‘STEM’ that can’t even identify good metrics for their designs.

steps off high horse atop a soapbox

Rereading this, it sounds like an indictment of the ‘engineering’ education we’re giving our students, but there’s many individuals and teams out there on CD that use and are pushing towards these more useful metrics like ‘sprint distance’. What are some ways your team is making more end-goal-oriented engineering decisions?

2

While this post is largely on the right track, it really comes across as kind of smug. It presents a lot of legitimate tradeoffs as cases where there is one Objectively Correct answer and the decisions people are making are Objectively Poor, and I think that sort of thinking here, while tempting, lacks the nuance and perspective of the whole engineering process.

Oftentimes designers think of FRC robots in terms of “optimal mechanical design”, as in, “this is how the machine should be built for absolute best performance in ideal conditions”. But this is focusing on only a small part of the actual engineering design problem. The challenge isn’t to build the most optimal robot, actually - it is to create a combination of a robot, a team, a drive team, a strategy, etc that maximizes that team’s competitive success given their constraints. When you analyze robot design decisions from this wider perspective, they often make more sense than you might expect.

Taking out weight has an opportunity cost. It ALWAYS does, in one form or another. Design effort, risk of failure (particularly with less knowledgeable designers), manufacturing time, restriction to future iteration, etc. And the opportunity benefit of a slightly lighter robot, while sometimes significant, is relatively marginal when you compare it to things like driver training, reliability, generally being done on time, etc. When you’re at the 85th percentile of FRC, that benefit gets a lot bigger, but when you’re at the 40th to 80th percentile, you haven’t maxed out the benefits of those other areas yet, and you aren’t exactly sitting around waiting for something else to put effort into during an FRC build season.

Optimizing sprint distances, to be frank, is a fringe thing. Teams struggling to complete every game objective have way better uses of their time than getting a 5-10% more optimal drivetrain speed or gearing. They have to optimize the whole problem - robot, drive team, strategy, etc. - and getting the design of the robot perfect is not always worth the effort versus improving other aspects of a competition team.

And there is SOME rationale to a heavy robot, particularly for “not traditionally good” teams - they have an expectation of potential scoring failure, and need to be able to play effective defense, and they feel weight enables them to do so. It’s not always (or imo, often) worth the tradeoff, but there isn’t a complete lack of logic there.

Of course, many designs are bad, poorly thought out, or just generally rushed in FRC - that’s the nature of a competition for high school kids where teams have varying access to knowledge, support, and resources. I’m just trying to say, it’s not really a matter of stupidity in a lot of cases. It’s not that you and the top tier are the only people who have figured out this is important - it’s that there’s just so much more in play than that in an FRC build season, and tradeoffs have to be made.

3

Eh?

4

Your in the weeds here in my opinion so Ill list some bullet points.

  1. Because for FRC it does not matter because…

  2. Not every team has the ability to make a fully functional robot no less the most efficient robot.

  3. In the real world of product or application constraints design choices are more flexible. For example the easiest and quickest way to reduce the weight of an FRC bot would be to ditch the Lead acid battery and go LIPO. no engineering knowledge necessary however safety outweighs the “just cuz”

  4. Lighter more efficient motors, compressors, control systems, bumper materials etc…

For good reason we have what we have but if the objective was to make the lightest bot these would be my target. Design is certainly important but show me a team who purposely targets making their robot as close to the 120lb limit as possible? None of the team do its just the result of the constraints we have to work within, and lack of experience in some cases.

5

6

7

Before everyone starts down the LIPO rabbit hole (too late), the battery isn’t part of the 120lbs.

8

Yea not a new concept…thus my reasoning for safety didn’t think I had to spell is out. That said I am sure you have a cell phone, laptop, or other device with them. Used in everything today.

9

It is, however, part of the mass we are trying to accelerate (up to design speed or across a design sprint distance). That’s what the discussion is about.

10

No but the OP was about lowering overall weight just for the sake of lowering overall weight. As you stated its not part of the weight allowance for FRC but again the OP stated why not 60 and not 120 so the rules are being challenged for the sake of a lighter robot.

11

First, I like that after a long rant about how we’re all making poor engineering decisions you acknowledged your “high horse atop a soapbox”.

Second, I think your question at the end here is the most important part of your post. Our goal is always to get our students to make goal-oriented engineering decisions - and if I could get every student on my team to consider every possible variable involved, that would be awesome, but that’s difficult to do within a six week timespan.

The reality is that we built an almost-120 lb robot this year because we determined that shooting was important to us, and we don’t have the time to optimize our weight. We used the same shifting gearboxes on a six wheel traction drive as we’ve used for 3 of the past 4 years because our high gear is “pretty fast”, and the low gear is pretty much traction limited, and that makes it so we can get a drivetrain done within the first week that functions pretty darned well.

When we designed our shooter, we went into much more engineering detail because we needed to, and the difference between a 70th and 95th percentile shooter is enormous. And we were afforded that time because we spent less time on the drivetrain.

When we designed our climber, we did some rough calculations of what we’d need to climb, then we powered it way more than that because we could.

The point is that I’d love to put tons of engineering time and effort into every system on the robot, but this is a program for high school students on a very limited timeframe. If my students understand why we made the decisions we made, I’m happy, even if their knowledge doesn’t extend to the differences between adhesion and friction or how to calculate a robot’s ideal sprint speed.

12

Disregarding the LIPO tangent…

That’s quite the soapbox you’re speaking from.

I’ll just echo what Chris stated - “When you’re at the 85th percentile of FRC, that benefit gets a lot bigger, but when you’re at the 40th to 80th percentile, you haven’t maxed out the benefits of those other areas yet, and you aren’t exactly sitting around waiting for something else to put effort into during an FRC build season.”

Those of us in the lower percentile are very aware of what we need to optimize prior to getting to the lowest weight possible.

Frankly, and I’m not taking a shot at your (former?) team here, but it looks like even 4213 has much to improve on before making this a priority.

As for “For everything else, the motor is accelerating and decelerating. You must look at a curve. VexPro even has those nicely for you.”

…I don’t know if the majority of folks on CD are the ones who need to see this.

13

OH wow, missed this first time.

Your two power comparisons are potentially true in some cases here, but are not universal truths in FRC.

I would agree that the general trend to push is 4 CIM and be as light as possible if you’re never going to play defense, but neither is a firm rule.

14

We tend to build light out of general habit. Then we often end up adding weight at the end. Not because we want to get to 120 lbs. But because we want to be able to locate the center of gravity properly. Some year’s that’s more important than others, but I find it a point of pride that my team’s robot has never, in the 11 year history of the team, tipped over during competition. Even when we fell off the top of the pyramid (multiple times) in 2013, we could just drive away. Even when it seemed everyone was tipping over on those bridges in 2012, we would just fall back on our wheels when our alliance tipped the bridge back towards us.

So that’s one reason a team could be building heavy.

The other reason we build heavier than needed? We over engineer the solution to ensure it doesn’t break. With more time and effort, we could do stress analysis and testing to figure out where the line is between constant fixing and periodic adjustments. But why spend that time and energy? We have the weight budget, we might as well use it to tackle the items highest on our priority list. Sure, lowering weight to increase mobility might be on the list, but it’s well below building an “unbreakable” robot. I’d rather have a durable robot that works every match than have one that is a little more maneuverable but spends a match or two in the pit because something critical broke.

15

You’re right. Just wanted to stop a LIPO debate before it got out of hand.

16

In defense of heavy robots:

  1. more material makes things stronger. Stronger things are less likely to break. I’ll take robust-ness over acceleration in most games. Crashing and breaking can put you out of many matches. Yes you can make a robot that can traverse ramparts and a moat with styrofoam wheels, but it’s just too risky. I’d like to win, but I care more that we get to play.

  2. Many teams don’t have facilities to machine an aluminum extrusion into a lattice work that weighs half as much for weight reduction alone. Many team use the kit frame, and yes it can be weight reduced, but we don’t have time and people for that.

  3. Robots that ‘do every thing’ have a way-better chance of getting picked by alliances. Yes, if your robot does gears amazingly, you can get noticed, but how many teams on Einstein didn’t DEFT this year? Skipping a mechanism is a risky thing for people that are considering your 'bot.

17

I’ll second this and what a lot of other people have said about pushing the limits of acceleration not necessarily being the priority. I think that potentially you’re bringing your experience in FSAE back to FRC and having new design priorities that aren’t the priorities of most teams. In FSAE one of the (if not the) main priorities is to go as fast as possible. That means putting a lot of effort into improving acceleration like you were talking about. But eking every second out of travel time isn’t a priority in FRC like it is in FSAE.

I would argue that an incredibly slow, heavy robot that could score a lot of points is better than a really fast robot that can’t score points at all. Obviously those are extreme examples and a fast robot that scores points is better than both of them. But I would argue that most average teams in FRC (mine included), if they tried to build a robot where their main priority was going as fast as possible and their second priority was scoring points would run out of time and end up with the second robot in my scenario.

18

In response:

  1. Never said my former team doesn’t have room for improvement. :stuck_out_tongue: Knowing what I know now I’d be pushing for this even more strongly were I back in a team.

  2. ‘not having the time’ isn’t a very good excuse for not trying to learn the fundamentals and principles. You don’t need to do every single stinking hand calc. Build/buy a gearbox that you can swap ratios out of, slap bricks on it, drive it. Time which setup hits a distance you like the fastest. This is a job you can have some newbies do during offseason and is miles of improvement over ‘18 ft/s because… speed’. (In fact… this method is better than even hand calculations because models are imperfect)

  3. If you don’t know where you should end up, you’re never going to get there. Being a new team or being low on resources or people is no excuse to not try to do good engineering.

  4. Some comment about weight rule… make the weight limit 200 pounds for all I care, my point was that lowering weight of your bot can be beneficial. Heck, make it 200 pounds- the benefits of saving weight will become that much more apparent.

  5. No FRC team is going to get to this level of nitpicking (if they do, kudos, and let me see all the FEA). But it seems silly to have this big high-budget competition come down to ‘fiddling around with expensive toys’.

  6. You are right, perfection will always be sacrificed at the altar of timeline. :smiley:

19

*in week 1/2/3. Comment was more about that if you’re building a gearbot, it shouldn’t weigh much at all. Take out the hole saws while the drivers are sleeping rather than enjoying their 4 weeks of drive practice.

20

Or use thinner material?