Best Practices for Modular Intakes?

No matter the year or what the game piece is, we’ve all seen some robot with a perfect intake that functions flawlessly and never misses a game piece. I’d like my team to get to that point, but we need some assistance.

Every time we’ve built an intake in the past, there was a ton of “build it halfway, decide to start over” due to student inexperience. I’d like to have them build a good “reference model” that both current & future students can take inspiration from, and have it be modular enough to quickly get a prototype up and running during build season.

As a summer project we’re looking into having our mechanisms group design and build a modular intake that can be adapted to most types of game pieces. What are some of the best intakes you’ve ever seen, and what are your team’s best practices for designing and building modular components like this? Any resources, photos, videos, etc are appreciated.

Thanks everybody! :smiley:

1 Like

Every game piece is special, and requires different geometry. There is no one size fits most for intakes.

A 2015 intake has no chance to work on 2016, 2017, or 2018 pieces. The thing that makes these amazing intakes so… amazing, is that they are precisely tuned to the exact geometry they want to intake, thats how you get a great intake.

There are too many possible combinations and ideas that one base model just wont work. Your best bet is to have a few different types of wheels around the shop (2" + 4" compliants, different durometers, fairlanes, colsons, etc), CAD up some quick geometry, and cut it on a band saw by hand or if you have one, a CNC. If you checkout 254’s build blog from 2017 here: https://www.team254.com/category/frc/2017-frc-build-season/page/2/

You can see just how much tuning they do to get it just right. Usually modularity is kinda a waste of time in FRC because the season is so short, you only have time to replace a few things, so your effort is best spent making things easy to replace or dissmble for various reasons.

Our intake this year was pretty awesome, we came up with a pretty good idea, tested it out of bandsawed wood, and it worked well so we tested a few different wheels, a polycarb CNCed version, and we were happy with it so we kept it all season. If you arent happy with how your intake is performing, go on youtube and search FRC, sort by newest, and during season tons of videos are being posted you can take some ideas from. You have to choose a date where you say “Its good enough”, and move on.

Compliance, Alignment and Grip.

Alignment -
To hold/move a game piece you need first align to it. In most cases wider is better, as a general rule the wider your intake the easier it will be to utilize. You can cheat a bit with CD7 style in takes (see a lot of 2016 robots for these) such that you can swing the intake in from the side to increase the effective width. Or with “vectored intake wheels” ie take some mecanums and stuff 'em on your intake. Make sure the game piece has a clear path not blocked by the rest of the intake.

Grip -
This is one place where your choice of wheels comes into play, but it also plays in with alignment. Obviously if you have a big gaping set of rollers it may have a wide area it can intake from but you give up on grip. A common approach here is to have actuated intake arms. These can be driven by a sensor or a driver control (preference is sensor if reliable). See a lot of 2015 intakes for this approach. There’s a lot of great information on intakes in the 2015 Behind the Design book (which came in your 2016 KoP) Material choice often comes into play here, some wheels grip certain materials better, having a wide selection of wheels to play with is beneficial.

Compliance -
You need compliance somewhere in the system. If the game piece is squishy (like 2016) you can rely on compliance there, if the game piece isn’t you need compliance somewhere in your intake. Again, going to 2015 (because it’s well documented in a book) is a good example. This year a lot of teams utilized squishy wheels in lieu of compliance elsewhere, the best tended to utilize it in both places to let them tune the exact amount of compliance.

By balancing all three of these factors you can tune an intake to the specific needs of the game.

That’s the “let’s sound scientific about it” explanation. Really though - when I see a game piece there’s just a library of intakes that I’ve been involved with over the years that I draw from. Need a box? 2015 style intake. Need a small ball (<1/4 the robot size)? over the bumper roller with grippy material. Large ball? 16 style beater bar.

The overall rule I use when looking at intakes is “touch it own it” the second the intake touches the game piece I want it to become mine without any action on the part of the driver.

1 Like

For 3538’s 2018 cube intake, it took loads of prototypes and revisions to get it near acceptable. In the first two weeks of build season we prototyped four or five different intake styles, then revised the best one multiple times more in the coming weeks.

When I was translating the final prototype into a CAD design for “final” fabrication, I intentionally isolated the arms of the intake from the baseplate to make the intake modular for further revision (also to be able to fit it on our CNC table, and to make assembly easier). Several geometric changes were made after fabrication of the first claw revision, taking advantage of the modular arm system to reuse the same baseplate over multiple revisions.

The CAD design itself was specifically done in a way that solely the critical dimensions of the intake (distance between wheels, diameter of wheels, etc) can be input, and SOLIDWORKS rebuilds the parts to conform to those critical dimensions. This allowed me to make small changes without having to redraw core intake structure every time.

I have an unfinished document detailing the design process of the intake here: https://docs.google.com/document/d/10iBqHpA-E1Ube6yqM2XBjiEwJh6mS6XQUWZx2FIph_I/edit?usp=sharing

There’s a bit of a resource issue with this as well. As the previous posts have indicated, slight changes in geometry make huge differences in the performance of an intake, and it’s really hard to verify the impact of a geometry change without actually building a prototype and testing. Having a quick turnaround on 2D parts in wood or polycarb lets you iterate much faster. Wood laser cutters are great for this. If you’re on more for a budget, you can print out a 1:1 drawing and do it on a vertical bandsaw.

Great intakes are great because they’ve gone through many, many iterations. Double digits are not uncommon.

The best practice I know for intakes is to make it hot-swappable, with a good rigid & robust interface to the rest of the 'bot. Easily accessible reusable fasteners & electrical connectors laid out in a sensible way FAR out weighs practice with specific geometries or wheel strategies.

Why is that so important? You’ll spend more time iterating the geometry and settings of your intake than pretty much any other part of the robot. You’ll probably swap on a new one on Practice Day of every Regional. The less time you spend taking it on and off the robot, the more time you can spend on literally everything else.

A little anecdotal evidence to make my point… On this year’s claw, we were 3/4 of the way there - four bolts held it in place, and the power connectors were dropped in and ziptied at the same “level” of subassembly, but the Talon data, arm encoder, and two cube proximity sensors were routed deep into the robot… redoing that routing every time burned valuable effort that could have been testing and iterating. And more importantly, when it came time to swap in Withholding Allowance improvements on each Practice Day of our regionals, we were burning 20+ minutes on rebuilds that should have been <5min tasks. :ahh:

Also +1 to CAD strategies with design/layout Sketch of critical geometry, driving the rest of the structure/non-critical geometry, for quick ez-pz improvements

Keeping this in mind was also a significant factor in 3538’s intake revisions this year. All components of the intake were installed with only 4 8-32 bolts that were the easiest thing to reach on the robot. At competition we were ready to swap the intake for a replacement in the turnaround time of any match.

This made it really easy for us to swap the actual competition intake between the competition bot and practice bot for tuning, identification of problems, and revision, using withholding allowance.

In terms of building a “modular” intake, I do like the concept of being able to remove the intake separate from your arm/elevator/manipulator when reasonable to do so. This does support a lot of ability to iterate and repair as necessary. This is something that we took advantage of greatly in 2015 and 2018, and to a lesser extent in 2017. There’s not really a “one size fits all” approach to this for making a removable intake, as it will vary. However, if your intake is rotating around a shaft (as ours was in 2017 and 2018), making that shaft easily accessible and removable is a great way of easing the possibility of a modular intake.

As for how we develop our intakes themselves, the weekly videos we’ve posted in build season the past two years shed more insight into that than I can put into a text post.




Finally, never stop iterating. Intakes are one of the best places to practice constant iteration and improvement, both during and after season. I lost count of how many different permutations of gear ratios, wheel quantities, and wheel placements we tried over the course of build and competition season.

As a general rule, side roller intakes work well for rectangles, and top/bottom rollers work well for gripping round stuff (especially balls). Side rollers have trouble gripping convex surfaces. We copied 1114’s intake from 2015 to make our intake, and with slightly more code testing it would have been perfect. As it was, it performed very well.
Iteration during the prototype stage was key. We messed around with the prototype before copying the dimensions directly into CAD.

Our very first intake prototype was the spare head for our 2011 robot, rotated sideways and mounted on some 8020. We worked from there by just playing with speed and geometry.

Extending my own reply - this extends to strategies around sensor placement & code as well! Encoders attached to the “mechanism position” as opposed to a “motor position” or “intermediate shaft position” might not need to be re-calibrated between swaps. Limit switches to automatically reset encoder boundary conditions work too!

It also includes code architecture - keep your constants as variables in a separate file, organized by subsystem, so that you can quickly find and edit those. No magic numbers hidden in the middle of a mess of subsystem controls!

This is a very good summary of what is needed for an intake. After both struggling with fuel in the 2017 season and watching some awesome gear ground pickups I studied until arriving at essentially these same three principles. One lesson that the comparison between 2016 and 2017 made clear was about where you can build compliance into your design:

2016 Boulder: Game piece compliant
2017 Gear: Wheels or intake structure (or both) compliant
2017 Fuel: Backing (hood material) compliant

All three methods work, but you cannot intake, convey, or shoot a game piece effectively, unless you have compliance somewhere in your design.

Based on this, we came into prototyping this year knowing that the power cubes wouldn’t compress, and that we would therefore need either compliant wheels, structure, or both. We ended up using AM 4" compliant wheels and pistons, which provided compliance in the structure, and allowed us to open up a very wide acquisition zone. Overall, we were pretty happy with our intake this season.

This.

We are now referring to these aspects as separate things after seeing intakes this year that had one characteristic but not the other. We also plan to require both for all future intakes. We say “touch it own it” to describe speed of collection and retention ability and “continuous” to describe the absence of operator action. Both of these combined make for an awesome intake. Continuous intakes can dramatically increase the speed of play and is likely the reason that 1678 specifically has a golden rule that calls for “rolly grabbers” meaning continuous intakes.

Finally, to comment on modularity, I agree with others who have said to make the intake a module that can be easily removed and replaced with a new or spare version. On bag day, we tested an improved version of control for our intake arm close to midnight. The mounting of our intake allowed us to take it off in just a couple minutes, which gave us the time we needed to test the code thoroughly. The good mounting also made re-attaching the arm at competition nice and quick.

1 Like

This.

I think the best way to fill your goal OP is to have a a couple vertical rollerbar intakes (E.G. 1619 this year, 254 ball intake 2017) and a horizontal wheel intake that can either contact the game piece with one wheel per side (195 this year) or more (254 this year, 118 in 2015). Make sure the horozontal intake has a widely adjustable width. Also make sure you can jerry rig the two types together into one mechanism. Who knows? That might just be the meta some year.
Echoing what others have said, the best intakes always come from tons of itteration and fine tuning for that specific game piece. But having those bases should give a solid start to your prototyping.

I want to point out that there is seldom one design to rule them all. There were three or four intakes I saw all over the place this year. 1114 2015-based designs with hilariously good ‘touch it own it’ and alignment ability; 118 2015-based designs cough254cough with excellent alignment, control, and speed; 148/118 2018-type designs with fast action, light weight, and mechanical simplicity; 33/3487-style four bars with very good flexibility, alignment, and control. We even saw some vertical rollerbar intakes (1619, 1678 early on). The point is, teams chose the intake that worked for their robot as a whole. Vertical rollerbars theoretically had the best ‘touch it own it’ capability, but that intake wouldn’t work for a flywheel shooter like 4481 because they needed to align the cube with the center of their robot for firing, which a rollerbar wouldn’t do.

‘The best intake’ is a matter of what design both works best for your robot and can be made within the season. (Of course, offseasons can be fun too.)

Adding to what others have said, I doubt you will find a truly “modular” or “universal” gripper for multiple games that applicable in FRC - it will make too many sacrifices for each game piece.

Modularity is useful however in limiting the scope of different subsystems and then deciding on one interface between that subsystem and the mating or adjacent subsystem. Just as an example, a 2018 intake could pivot around an axle that will not change between separate iterations either the intake or elevator.

Being able to quickly swap intakes is a plus (or a requirement), but when designing/iterating, forcing your team to stick with one interface will ease the integration process between all the subsystems greatly. Obviously the interface can change if that is required, but it allows you to deploy new versions, revert to old versions, or create backups within a much more streamlined process.

148 inspired me this year - I think they were on their Mk.9 intake by Houston. Just awesome. No clue if they approach the design like this, but it is how I would structure the overall system to achieve the huge amounts of iterations they did.

(also keep in mind electrical bits here - extra connectors can save so much time)

There’s a lot of great discussion here so far, I appreciate all the feedback!

As a couple people noted, a modular design will be helpful for initial prototyping, but not an optimized final design. I agree, there’s a lot of room for optimization once a base design is selected. Right now we’re just trying to get away from just using the first prototype AS the final part (because the team started from scratch each time).

Regarding “Alignment, Grip, & Compliance”, I really like this breakdown of the important design concepts and plan on explaining harvesting mechanisms using these ideas. Same with “touch it, own it” and continuous operation.

I like the idea of hot-swappable mechanisms (for both mechanical connections and wiring) to get upgrades installed quickly mid-season, we just have to figure out a standard interface… we always send new kids through bandsaw and drill press training, so that would be a good opportunity to practice making a bunch of standardized interface plates (suggestions welcome!)

The code portions are further out from my wheelhouse so I’ll have to get with the programmers to see what they think of standardizing our mechanism inputs.

Not necessarily needed on final designs but having adjustments built into designs can make prototyping easier. With intakes small changes can have large impacts.

I Know what CD7 intakes are but where did they come from and why are they called that?

By C7 I assume you mean CD7…

They come from Team 47 - Chief Delphi (you might recognize the name) from their 2002 robot. https://www.chiefdelphi.com/media/photos/14307

That’s pretty interesting.

To expand on this topic…what are some other examples of either this style of intake, or totally different intakes that seemed “just crazy enough to work”?

As mentioned, the CD7 intake was fairly popular in 2016.

I know 27 built it in 2006 and again in 2008 (yes it’s giant and glorious)

I believe 973 has used it a few years with reasonable success as well.

Some crazy enough to work intakes - 125’s mecanum intake from 2014 allowed them to grab a ball off center and have it JUST WORK. A similar approach was applied to the 2016 intake. I know there were a few other teams who did similar, I just didn’t work with them.

I’ve always been very fond of the 16 style lexan beater bars (2008, 2014), 558 also made great use of one in 2014 as did 1678 (only they made use of 2) I repurposed the motor in roller concept for 79’s intake though that was more rigid than the other two (still fairly floppy though). The primary benefit of these intakes is that they can take a lot of nasty hits and flex out of the way.

As an example of how there’s exceptions to all rules - 217’s giant flat gripper in 2011 is one of the few examples of pinch grippers that were incredibly effective.

16’s lean box in 2012 was awful effective as a multipurpose manipulator/intake. For another great example see 67’s utility arm in that year.

971 has had a series of absolutely bonkers intakes that I’m not even going to go into.

And if you want just pure magic, 469’s 2013 intake system works in mysterious ways.

Just to clear things up, when most people say “touch-it own-it” they are referring to what you call “continuous”. Touch-it own-it is derived from the intaking process being complete without any human intervention beyond touching the game piece to the manipulator - hence the name.

In terms of speed / retention ability, I don’t know if there’s a specific term for that, that’s just what an intake is supposed to do. “Intake speed” and “holding ability” or something, I guess.