On Tuesday February 19th, the Robowranglers “shipped” our 28th robot!
At this point most of you have probably seen the robot, but as my faithful and dear readers (and anyone with a calendar) can note, it’s been weeks since my last update. How did the team progress in that time? How did our process unfold?
This season has been a bit of a struggle since the beginning, and the last few weeks weren’t any different. In talking to friends and seeing the comments folks are making: it seems like we aren’t alone in this struggle, and that lots of teams had similarly “weird” seasons. Let me show you how we finished out. Then we can get into he details of “the robot”.
To skip ahead and get more information on Overhang, check out the post 2019 Update 6: Overhang.
“Wow, JVN hasn’t posted in a while… I guess the Robowranglers are really far behind?”
While that’s true, you shouldn’t use my personal motivation to write blog posts as a metric for Robowrangler progress. I’ve been personally busy and unfortunately, documenting our robot build has taken a backseat to “make sure our robot actually gets built” but more importantly blogging takes a back set to “make sure the companies I’m supposed to be running are actually still running.”
But again… we ARE behind. We’ve been playing catchup since very early in the season. Thankfully, on the Robowranglers the end of Build Season is really just the beginning. Even if the robot have isn’t “done” there is plenty of time to rebuild it. Even if we ARE happy with the robot we have, we’re still probably going to improve it. Lots of people are already making jokes about “well this wouldn’t be the first time we rebuilt a robot during the competition season.”
My recent guidance to the team sums it up:
“Trust the Process but… GET TO WORK!”
The Definition of “DONE”
As someone who writes a lot about iteration, I have a literal collection of cliches. One of my favorites is that ‘DONE’ is a 4-letter word. In the FIRST Robotics Competition teams have opportunities to continue to improve their design even after build season. This raises an interesting question.
When should we unveil our “finished” robot?
The Robowranglers have been doing robot unveil videos for years, and each year it seems like the robot we compete with at the Championship gets more and more different than the one we “unveiled”. (Our 2017 full-rebuild being a great example of this).
This season Texas has gone to the “district model”. Which gives us even more opportunities for iteration. Now we can unbag our robot for up to 6 hours before each event to do modifications. With the 148 manufacturing capabilities… those 6 hours open up LOTS of opportunities.
Did this unbag time influence our season
Did we behave differently knowing we have a safety net beneath us?
We bagged a robot. We unveiled a robot. That robot is going to look very different when we compete at Amarillo on March 8th. It’s going to change again when we compete in Dallas on March 29th. (Honestly… the practice robot already looks different than what it looked like in the unveil video.)
Next year, FIRST has announced “no more bag”. This means, we won’t even have the artificial deadline to limit our iterations. We’ll be improving the robot slowly and continuously from kickoff until competition. (This sounds so incredible. Just typing it out makes me smile. It’s going to be sooooo much less work.)
Next year with no bag day, will the Robowranglers even do a robot unveil?
We think that this might be our last unveil video. We weren’t even sure if we should do it this year, knowing that the robot is still evolving. What did this evolution look like in the run up to bag?
The Drivetrain Change
I’ve learned a lot of lessons over the years. Most of those lessons came the hard-way… from making mistakes. I try not to learn the same lessons twice. Unfortunately, I’m starting to get forgetful!
The original drivetrain utilized dead-axles for the wheels, and dead axles for intermediate gear stages with belts syncing between them. The dead axles rely on bolt holes for position, and aren’t the highest tolerance. This isn’t usually the biggest deal, but the belts pull the gear mesh apart and everything kind of runs like crap. It takes a TON of work to get everything running “happy” and that work needs to be repeated each time a wheel is removed for maintenance.
We learned this lesson in 2014. Why did we make the same mistake again? The typical Robowrangler answer would be: “because we’re dumb.”
After fighting with it for a week, we decided to remake the drive plates and swap everything to live axles. Problem solved. “Good news kids, we get to iterate!”
The Intake Redesign
Very recently, our robot had a disc floor intake on it. It worked great. Pickup disc, score Disc. Pickup disc, pass disc to beak, score disc.
What’s the problem?
One: we didn’t realize how much it would interfere with picking up balls out of the depot. We like our “slurpy ball cycles” and it was very frustrating the first time our driver Josh went for a ball and the disc intake smacked against the steel bar.
Two: we’re not sure if the disc intake is as important as we thought. After seeing more teams unveil, after doing much more simulated match play, and after running through a few ‘war games’ among our strategists… we don’t think we need it.
Will it go back on? Maybe.
Will we wish we had it? Maybe.
Are we better at other aspects of the game without it on the robot? Definitely.
Did we spend a bunch of time learning about this and then redesigning? Oh yeaaaaah.
The ball intake in the unveil video is the new simplified version. Turns out once you remove a bunch of constraints relayed to disc intake integration… things get way simpler and way better!
Many people in the community were shocked to see our 2019 elevator was not the same round-tube construction we used in 2018. Why did we switch?
Using the new 0.050” wall extrusion from VEXpro, the weight penalty for the change wasn’t that bad. It also enabled us to design this elevator for higher loads than last year’s (at various points in the build we’ve had some ideas which might have needed this). Square tube is also a HECK of a lot easier to mount to.
That is my favorite gif of the elevator. I have no idea what Brandon is doing…
The gearbox for the elevator is designed with 10 different gearing options. We had the slow “high torque” option in there for the unveiling, but we’ve been tweaking it since then.
It’s just a smooth as the round tube. It can be just as fast (but we don’t think it needs to be THAT fast for this game). It’s only a little heavier. But… it has higher load capacity if we need it. Is easier to mount on, and is a lot more modular for robot iteration and maintenance. (Changing out an intake in 2018 was not a wonderful experience).
End Game Constant Evolution
The end game mechanism has had lofty goals from the beginning, and the implementation has been evolving for weeks.
You can imagine the integration work that went into it. We did some redesign work to shift the CG of the robot. This required us to re-tune the drivetrain handling.
Yes, the end-game mechanism works. Yes, it can do it “after the buzzer”. Yes, it lifts itself fully up and doesn’t touch the level 1 platform. Yes, it works “every time”.
No, we’re not happy with it.
Not because it doesn’t work, but because our priorities have shifted a little bit…
The Missing Mechanism
We’ve still got one mechanism in our system plan which hasn’t been added yet. Maybe for Amarillo? Maybe later? I guess we’ll see if we get around to it…
Cycle Tuning & Driver Practice
Every year at some point the robot gets together enough for a “full scoring test”. This isn’t as fun as you’d think. We power up the robot and run some scoring drills. Typically within the first few minutes we find something we want to tweak. “Pull it off the field, reduce the wheel compression.”
This goes on for hours. The drive team, and a few pit crew members making fast iterations to the robot. We usually make weeks of progress in these few hours.
Typically this happens sometime during week 4 or week 5. This year, it didn’t happen until week 6. While we made the usual progress, it was frustrating to be “late”.
Does this matter? No. We still got the robot ready for video debut. We still have plenty of time left to iterate before competition.
Is it still frustrating? Yep. No one likes being behind schedule, even if there are perfectly viable fall-back schedules.
After documenting our 2018 season, I received lots of comments saying something like: “It’s great that you share how you guys struggle. I’m glad our students get to see 148 fail. It makes you that much more of an inspiration.”
One: I’m genuinely glad we’re inspiring. That’s a big part of why I do this. I called the blog “FAIL HARDER” for a reason.
Two: Can we all acknowledge how funny that is? “Seeing the Robowranglers Fail is Inspirational!” Well… I hope we’re a lot less inspirational in the future.
“We celebrate failures as opportunities for learning… and I am getting SICK and TIRED of celebrating so much.” - JVN, 2017
Are we Dumb?
At one point, the battery was going to translate to dramatically shift the CG of the robot during the end game. I’m so soooo sad this didn’t make it on the robot.
Yes, this was my design. No, no one else on the team even humored me by pretending it was a good idea. JVN sad face.