Looking Back: 3 Day Robots

[cadandcookies]

Originally Posted by Siri

Personally, I agree with you--I'm a huge fan of Ri3D. But what non-anecdotal evidence could possibly be offered for either case? It's not like we're taking a census here. Would you be looking for something in particular?

I'm just looking for larger scale or aggregate data-- ideally in a statistically valid form. I don't really expect anyone (either for or against Ri3D/BB) to bring in data, but I think that's the only way to move beyond rhetoric (where the conversation is now).

As for metrics, there are a couple of higher level ways (that don't require new data) to look at-- all of which have their own problems-- rookie retention/overall retention versus prior years (problem: is affected by other factors like quality of game and uncontrollable factors), lower quartile scoring ability (doesn't necessarily mention inspiration or thinking, just competitiveness).

I'd be curious if anyone has an idea of how to analyze the problem while moving away from rhetoric and into more objective analysis (which I realize is difficult and unlikely to actually happen).

[BigBen]

Overall I would say it is a positive engineering source of information, but I think they shouldn't release all of their info within the first week of build when everyone should be brainstorming. That takes SO much away from the actual design aspect giving the easy out of "hey this works so lets do this" instead of trying to come up with something. A little research on past year robots with similar games would provide some similar amount of info, but at least they would need to work at it.

One of the other mentors on our team actually thinks they Ri3D robots should be used by FIRST to actually play test the game to work out the kinks in the system, would be much smoother early week competitionss if this was so.

In hind sight, I think having this resource was nice, but I think it shouldn't be made available until the end of week 2 of the build season. My 2 cents anyway.

[Andrew Schreiber]

Originally Posted by cadandcookies

I'm just looking for larger scale or aggregate data-- ideally in a statistically valid form. I don't really expect anyone (either for or against Ri3D/BB) to bring in data, but I think that's the only way to move beyond rhetoric (where the conversation is now).

As for metrics, there are a couple of higher level ways (that don't require new data) to look at-- all of which have their own problems-- rookie retention/overall retention versus prior years (problem: is affected by other factors like quality of game and uncontrollable factors), lower quartile scoring ability (doesn't necessarily mention inspiration or thinking, just competitiveness).

I'd be curious if anyone has an idea of how to analyze the problem while moving away from rhetoric and into more objective analysis (which I realize is difficult and unlikely to actually happen).

The big issue is isolating the variables.

We could pull retention data but that would be tainted by various grants coming and going last year as well as changes in focus. And it's impossible to say with any real certainty if a team would have continued without success. (or define success for that matter)

We could pull times the game objective was achieved and compare that to similar years. 2012 work pretty well from a quick glance but has some serious problems at a deeper one. Perhaps 2006 is a good data point? Too old? Idk

Or maybe we could divide teams into historical tiers and see if they've had a pair of better years the last two compared to their traditional performance. But then how do we factor in that event performance IS a zero sum game, someone has to lose for you to win after all...

Sorry bro, don't think I can give you the data you want. Don't think anyone can. Objectively measuring the impact of Ri3D and its ilk is simply not possible. Subjectively we all have to determine if we should encourage our students to investigate these challenges. But I think that'll be a student by student question for me. Why do we have to have a definitive answer anyway?

[cadandcookies]

Originally Posted by Andrew Schreiber

Sorry bro, don't think I can give you the data you want. Don't think anyone can. Objectively measuring the impact of Ri3D and its ilk is simply not possible. Subjectively we all have to determine if we should encourage our students to investigate these challenges. But I think that'll be a student by student question for me. Why do we have to have a definitive answer anyway?

Like I said, I didn't really expect it to happen. The search shall continue.

[Abhishek R]

I came into the competition season expecting to see a bunch ofBB/Ri3D clones, and so many so that it would make me dislike it. That there would be 4 designs, and teams would just pick and choose.

Instead I saw a few teams who did follow one of the paths, but they can't have done it without having gained some technical knowledge along their way and/or putting their own twist on the base of the design, whether it be a catcher, new drivetrain experiment, different power system for a catapult, or other function. And all this while, teams still found unique designs (i.e 118, 254, 1678, etc.) and still stand out.

I would like to see them return next year.

[tim-tim]

Originally Posted by BigBen

[snip]
I think they shouldn't release all of their info within the first week of build when everyone should be brainstorming. That takes SO much away from the actual design aspect
[snip]

While we are on the topic of design process and brainstorming, I will add a few of my thoughts. There is a great read out there Product Design and Development by Ulrich and Eppinger that provides a lot of insight into the design process. They list the Concept Generation, or brainstorming, portion of the design process as a five-step methofology:

1. Clarify the Problem
2. Search Externally
3. Search Internally
4. Explore Systematically
5. Reflect on the Solutions and the Process

A brief description of searching externally

External search is aimed at finding existing solutions to both the overall problem and to the subproblems identified during the problem clarification step.

One of the last subtopics in external search section is "Benchmark Related Products".

I do agree that it is important for teams to go through the proper steps before starting the brainstorming process (how are you going to play the game, what is the team's goal for what the robot has to do, prioritizing, etc.).

I am a fan of the 72 hour builds. I learned a lot by watching them, I can't say that we took what any of them did and placed it on our robot; but we did have some inspiration along the way.

[Xavbro]

So I was lucky enough to be apart of Robot in 3 Days AND the actual FRC build season and here's my take on it:

From the stand point of the FRC build season:
My team was watching all the BB/Ri3D builds and the approach they took was to "look but not watch." What I mean by that is they looked at the builds but didn't take intensive notes. The students all had something they liked from the different builds but I always asked, "what could be done to make it better?" Our team started with the choo-choo but realized that it was very limiting as opposed to a pneumatic launcher. We liked the JVN collector and stuck with it only after we explored all the possible options. Even though we had several robots done in 3 days to "copy", we didn't finish our design until almost week 3. The BB/Ri3D builds overall helped our team by cutting down a lot of prototyping that we would normally do. We could easily look at a video of a mechanism that we were considering and see if it gave us the result we wanted.

All in all, I comes down to how you let the build influence your students. You can't control how they feel about it but you can force them to think it through by making sure they cover all of their bases on why they want that design for the robot.

From the stand point of a Ri3D build:
Going in, Team O-Ryon's purpose was to build a robot that could complete the simple game objectives and at the same time, be something a rookie team or a team that was low on resources could build and still be competitive. We knew that we had a lot of people watching and that whatever we build could and would be copied. We also wanted to make it difficult for teams to just sit down and copy it bolt for bolt, screw for screw. That's why we didn't CAD it and release any CAD. We got numerous emails and messages about when were going to release CAD or if we could CAD something for them and as tempting as it was, we didn't. We answered all questions but told teams to improvise because we knew our robot wasn't the best and that teams could make something better with 6 weeks to build.

Looking back, I'm glad we did the 3 day build because we met our goal. Teams built our robot (some carbon copies) and were able to be competitive at their respective regionals, some even winning. At the same time, the veteran teams were able to take what we did and iterate and take it to the next level. That was the main goal of Ri3D when it all started last year. Not to build a robot FOR a team but to build a robot that a team could see complete the game challenge in week one and to give them a foundation to start with so that the overall competition level at all regionals could be more leveled thus getting all the students a better overall experience of FRC.

[themccannman]

Saying that ri3d reduced variation in robot design is entirely confirmation bias. I think a lot of people are greatly overestimating how much design variation there has been in past years. Aside from 2013 I can't think of a single year that had as many truly unique robot designs as this year.

[cstelter]

Team 3018 (Nordic Storm) was inspired by last year's robot in 3 days-- not their design, but rather the challenge of doing such in 3 days.

This year the students tried to build a fully functioning prototype robot and did a pretty respectable job of it. Their final design was largely based on their 3 day prototype. They used the drive train from our rebound rumble bot and focused on the manipulator and collector during the build, but the prototype was able to nominally do all aspects of the challenge that our final version could do.

3 day prototype:
https://www.youtube.com/watch?v=sE9H627YZyM

Final:
https://www.youtube.com/watch?v=y8sxcxqOOMA

It didn't hurt that we are a community based team and we had snow days Monday and Tuesday so they were able to build from Saturday@noon until Tuesday@noon. The original plan was Sat Noon-late, Sun 1-9, Mon 4-9, Tue 4-9 for roughly 30h or build time over a ~80hr period, but they worked pretty much all the waking hours Sat Noon-late, Sunday 1pm - Tue noon.

In the end, I think the decision to try to not just prototype ideas but prototype a complete functioning robot took us further faster than in years past. It certainly was a lot of fun and we wouldn't have considered it if it weren't for last year's Ri3d effort.