Designing custom frames

I have been working on the CAD for our robot in preparation for the extended season next year and I noticed that a lot of older teams have components with these intricate, hollowed out frames. Is there a certain mathematical process behind figuring out the size and shape of the hollowed-out pieces or is it just something that comes with experience/ intuition ?

This question may seem confusing so I have included pictures:

I would also appreciate any critiques to the design of our current robot (it’s pretty bad):

Are you referring to the lightening patterns (generally the tessallated shapes)?

Yes, I think so.

The hollowing out of the frames is usually referred to as pocketing, or lightening. Usually, when we lighten parts, there isn’t a strictly mathematical way we do it – rather, we have some understanding of which sections of a part contribute least to the strength of the part in the direction we care about, and then remove material that way in some visually appealing pattern.

Generally, this means preserving material near edges, not heavily pocketing things that take lots of impacts, etc, but really it depends a lot on your specific design. I generally visualize it by trying to imagine how a part may collapse if it was made of a much weaker material (like if our intake was made of paper and we ran it into a wall, how would it collapse?). Then, I try to visualize where supports would be necessary to prevent it from collapsing in that direction.

As a general rule, though, spending tons of time pocketing thin wall material (0.09" and thinner) is a waste of time just because it saves very little weight. Pocketing thicker material is often worthwhile for weight savings, but reducing wall/material thickness will save more weight for far less effort (switching from 0.125" box tubing to 0.0625" is pretty much a 50% reduction in weight, which you would be hard pressed to get with just pocketing). Where pocketing most often makes sense is when you need the thicker material for rigidity (gearbox plates, for example, are thick so that the plate doesn’t flex at all, resulting in fixed spacing for ideal gear mesh).

I just want to say, you don’t need to qualify your CAD by calling it pretty bad. On the contrary, it’s actually a pretty good design, and very similar to what we have. Even if it was terrible, you’re learning, and that’s all that really matters anyway :slight_smile:


I see, thank you!

Not sure what you are using for your CAD, but for OnShape there are things called “featurescripts” which automate certain tasks, (I assume other CADs have something similar). Anyway, One we use is one that automates these pocketing patterns. You basically just lay out a set of lines where the “bridges” go and it takes care of all the extruding and filleting. A great time saver.

1 Like

By hollowed out shapes are you referring to the triangle holes in the first one by chance?

Yes, the lightening patterns (as I have just been informed).

I am not sure if fusion 360 has a similar feature, but I shall look into it!

In the real world this is part of larger topic called “Topology Optimization.” Actual topology optimization is super overkill for FRC. Loads in FRC are generally not sufficiently understood / well defined to even set up a valid/correct topology optimization problem. Classic garbage in, garbage out problem.

If you are concerned about weight your team will be best served by focusing on places to reduce material thickness as @plusparth said.


More often than not, in FRC, it’s for looks as much as weight savings. Definitely not mathematically optimized at all.


As a point of advice on your current robot design, which is much better than you give it credit for :), I’d HIGHLY recommend trying to pursue a climber. Barring any rule changes, it is the single biggest way I can think of improving any robot that doesn’t have one.

Generally I’d say its better to steer away from pocketing because of the time cost. With not having bag I’d say its more viable now (not that it wasn’t before), but ultimately having less machining operations is better. In a drivetrain with extrusions, you could go solid and thin instead of pocketed and thick. Faster to manufacture and can perform similarly if not better. I’d go for simple circular holes if you are pocketing: quick to design and manufacture. Simbots did a great video on this. Greybots also made a great video on pocketing gearbox plates. If you have the manufacturing capability, you can also look into pocketing gears, especially larger ones. Would require a fixture and either a CNC machine or a drill press and some good 'ol ingenuity :stuck_out_tongue_winking_eye:

1 Like

In FRC, pocketing patterns are those that make the most sense. There is a thread on CD with a discussion about it from a while ago. FRC pocketing My general understanding is that the pocketing does not have to be as “rigid” as in the industrial practices, and so going with a reasonable pattern will also make sense time wise.

If you would like to look into the mathematical optimization of it, then have a look at this blog (It actually uses FRC robots as examples)

Also fyi, Solidworks has the vent feature that will create a ‘ventilation’-style pattern.
Inventor has the ‘mathematical’ optimization tool built into it.
In Onshape, there is the Featurescript Lighten that will help make pocketing patterns easier.

1 Like

I agree, real pocketing is not worth doing unless you have CNC capability. Otherwise, you probably could just drill a few holes for motor vents and maybe lightening.

We don’t go thinner than 1/4" on gearbox plates because that’s the thickness of the common bearings. Then we pocket for cooling airflow, weight savings, and aesthetic coolness (but we have CNC). Other teams make different choices.

1 Like

Would you recommend a soon-to-be third year team invest in a cnc machine? Our school has one but they refuse to let the robotics team touch anything.

Our team really wanted to implement a climbing system. I originally designed a single stage elevator for it but our team was already $500 in the red at that point. We got two pneumatic cylinders with 12 in strokes from one of our local teams but I’m still trying to work out where that would fit on the robot.

Hard to say without knowing what you already have, and what your overall competencies are. There are a number of threads here on CD about how to spend your first $X dollars, one is

I would first try to talk to the school and see if you can start working with it. With the right capability, a CNC machine is powerful. It opens up a lot of doors to different designs you previously wouldn’t have been able to use. Helps save time manufacturing when you can do it in-house. A small hobby machine can help with making gussets, or with some fixtures, machining extrusions.

I see. Well, if you ever need any assistance with the pnuematics climb route, feel free to ask me any questions. Our team ran two giant cylinders as our climber this season and we learned a lot about pnuematics calculations and air volume necessary.