As a student, I’m always trying to improve in my design skills and intuition, and others have said that I’m a “good” designer, but I’ve seen other designers that almost seem magical in terms of their experience and intuition - like being able to design mechanisms in a few hours or robots that work great while having an intuition of simplicity on how to accomplish game objectives in an elegant way. Sometimes the solutions don’t seem the most “cool” but I think the “simpleness” and execution of some of these robots amaze me.
There’s also design mentors, who somehow are able to spot out issues in designs with a quick glance, and have the experience to review and guide students to build some insane mechanisms. How do you start bridging that gap and hit that point? I have no clue.
I mean there’s always the stuff like design more robots, look at more robot cad, talk to people online, look at robots IRL, get design feedback etc but is there anything else? Is there a point where the only other thing is to just build a ton of physical robots to see where the actual problems turn out?
There’s no substitute for experience… And having failed, or learned from other’s failures. I’ve definitely logged my share of failures!
Exploring how others have designed robots is one tool that doesn’t involve building it. One caveat: CAD and reality rarely match. You will find lots of problems that got fixed in the physical world but still exist in the CAD.
You might look for opportunities to design stuff and build it. Maybe sub-systems and partial objects? 3DP parts to explore fit and function with low barrier to entry.
I would like my kids to explore arms driven in and out with a dual sprocket; I loved Steam Punk’s design. The carbon fiber deployable structure/arm was mind blowing and clearly needs exploration/testing. I’m still noodling on some more exotic stuff that’s not ready to be exposed to the kiddos for reaching with tape measures and for stabilizing tall robots.
There are books of mechanisms that are fun reading for mechanical types. They add tools and solutions to your mental tool kit!
Attention to detail goes a long way. There’s a lot of little details that can cause problems when failed to be accounted for. I tended to evaluate every little detail in CAD before manufacturing the robot for potential improvements or failures. The nice thing about CAD is that things can be changed quickly. This means that by the time the design is ready for manufacturing, most possible improvement can be implemented and potential failures be mitigated. Improving stiffness or reducing backlash go a long way as they can subtly eat away performance without being super obvious most of the time.
I’ve also found that microoptimizations do go a long way actually. I was never afraid to make that part a few ounces lighter, a half inch smaller, or whatever else can be done to push the part as close to perfection as possible. Improving a bot 1% 100 times is a large improvement and usually is what separates the good from the best. At some point, chasing these gains becomes natural and will result in much better designs regardless of any other circumstances.
Who are you trying to compare yourself to? Many mentors have many years of FRC experience AND an engineering degree - it is just totally unreasonable to expect a high school student with 1-2 years of experience to be better than them. Imagine yourself today, versus yourself when you joined FRC - you’re probably already much better than before!
But in general, I think these are some key factors that allow designers to be successful:
Experience with similar mechanisms. Has the designer built a double-jointed arm before? That helps!
Wide exposure to and knowledge of past solutions. Walking the pits, talking to teams, and seeing different solutions for 5+ years is extremely useful in knowing what works and what doesn’t
Deep understanding of the COTS catalog. Knowing that REV has the perfect part for thing X is a huge advantage
Understanding of physics and engineering principles. I did in fact learn things in college
Putting in a lot more work than is visible on the surface. Great teams have large student bodies, tons of resources, experienced mentors, and put in long hours. Students sometimes think about FRC designs during class, mentors sometimes do it during work. A mentor might suggest something on the fly but have thought about it for an hour during a 9am budget meeting the day before
Besides just plain old experience, designing parts with parameters/variables can go a long way in making things go together quicker if you aren’t already using them. It’s a lot easier to change the frame dimensions multiple times if you just need to change a number vs needing to modify each part individually.
Another that helps alot is getting hands on experience building parts of the robot. It’s alot easier to figure out how easy it is to build something if you know how it is going to be built.
This is a really good response that summerizes a lot of my thoughts on this topic.
I want to highlight something else: it’s a fools errand to approach design solo. Bounce your ideas off other people. Have them review your work. Design reviews from people with knowledge and experience helps tremendously with getting things right on the first time and for me is an integral part of the design process.
Regardless of what you do, know yourself and seek self improvement. We can have all the shortcuts in life…it will never make us a better person.
The fact that you are asking this question shows that you are better than good in what you do. When you stop asking questions, that is when you have given up.
I have more thoughts on this, but I’m pretty wiped from Champs still so I’ll just add a little bit here.
This is a balance. Things that work in CAD will not necessarily work IRL, and realistically the first thing you build won’t ever work the way you want. The earlier the design, the larger the changes you’re going to want to make. If you want to be time efficient, start with a coarse “good enough” cad model, build it, and start making a list of necessary tweaks and changes. Then go back and add those details to cad and build that. The more of those cycles you can get through, the better the final product will come out.
If you spend a bunch of time optimizing your first design before building it, you will end up taking longer, solving problems you didn’t need to, and adding unnecessary detail and complication to the final design. The hard part is defining “good enough for a first draft,” and that comes with experience.
A good designer is able to create complex systems/robots while a great designer is able to create simple systems that perform complex tasks.
For me system and tasks integration is what makes the difference and as in everything in life nothing will replace experience, research, and the ability to fail forward and be in a never ending learning cycle.
Something Ben Martin said one time stuck with me. I don’t have the exact quote, he said there is FRC domain knowledge, and there is general design/engineering knowledge… Between the two, FRC domain knowledge is a bit more important toward having competitive FRC success, but of course you need some of both.
You see this every year as students with maybe one physics class under their belt and who learned CAD the year before but who steeped themselves in FRC to gain deep FRC knowledge design highly competitive robots.
To me, the very strongest designs seem to come from teams with personnel who have EXTRA deep FRC knowledge/experience and who also have deep general design/engineering knowledge. I think the two things together enable them to see design opportunities (including concepts to prototype) that others don’t. And the general design/engineering knowledge may enable some sophistication that translates to an extra 10-20% efficiency over what a primarily FRC knowledgeable team could do. Plus teams like that tend to also put it a lot of time, magnifying their differentiation.
Do you think this is true for this year? Take 2056 and 1323, two highly competitive robots with vastly different designs. 1323 was clearly the most dominant this year, with like 3x of the mechanisms 2056 had. I feel 1323 designed a complex robot to do a moderately simple task, while 2056 had the polar opposite approach. Do you think 1323 had a great designer behind it?
This cropped up in interesting places for us this year. When we had to scramble for some plates to mount a cube pooper, I first spammed down a 1/2" pitch hole pattern. I then thought about it and changed it to a 1" pitch, because the brackets holding the pooper itself had a 1/2" pitch and would give us plenty of adjustment. That saved us dozens of holes on the router, which meant we had first (and, it turns out, final) parts ready to clean up and mount appreciably sooner (especially tangible since this was the Wednesday before our first event).
I’m looking forward to being able to churn through more iterations faster next season as our resources and skill sets grow.
I think this is an example of (overall) design vs execution. There are plenty of different designs with different pros and cons. In my opinion some overall designs are “better” than others but what is equally if not more important is the execution of those designs. Both 1323 and 2056 are teams that can execute their design to the highest degree. This means that regardless of design they’ll be competitive but because they also picked “good” overall designs they’re even more competitive. One thing I liked about this game was that there were multiple great designs that were extremely competitive instead of all the top robots looking more or less the same.
Every year there are some teams who pick a seemingly strange or none optimal design but they execute it at a high level and consistently outperform teams who picked a “better” design but didn’t execute it as well. As you get to higher levels of competition you might see these robots struggle because while they have a good execution of their design, it may not be able to reach as high of a potential as other “better” designs.
Keep in mind terms like “good, bad, and better” are pretty subjective and loose when it comes to most years.
Good designers understand what factors contribute to reliability, repairability, and effectiveness.
Great designers know what factors DON’T contribute to those qualities.
What I mean is that FRC is an extremely limited environment for most teams, you only have so much time, money, and manpower. Knowing what will be the best use of those limited resources and designing a system that optimizes for those qualities is the most important skill a designer can have.
So (aside from looking at successful robots) my biggest piece of advice for anyone who wants to get good at designing in FRC is to get really really familiar with your team’s manufacturing capabilities.
I think both were great designs this year but had vastly different design goals. 1323’s design was heavily influenced by the cube passthrough system to cycle hybrids quickly combined with being able to extend to the double substation and have bumpers hit the same time as the wall (similar to 254), which led to extra DOFs being necessary. 2056 instead tried to limit DOFs as much as possible (2 was the minimal this year for placement unless you were shooting cones) and didn’t have the same double substation intaking issue that 1323 did since they prioritized ground cones and single substation drops for pickup.
