“Do you guys have a good plan? What do you think the robot is going to look like?”
I know most people won’t believe the answer, but if a Robowrangler tells you “We don’t know yet” they’re telling the truth. As we cruise into Week 2 of the build season it feels like we’ve raised more questions than we’ve answered.
Each day we know more than we did the day before.
Each day we know more about “what we DON’T know.”
However, I think we’re finally honing in on a strong direction for the system and a set of design team priorities for the detailed design.
The robot pictured above is “Devastator2019”. (Named for the Decepticon which combines from multiple other Transformers, not for the Cheesy Poof robot of the same name).
It combines my favorite features from a few past Robowrangler robots.
Seems like a winning concept, don’t you think?
The community seems to think it’s a winner.
Is this what it feels like to be a rich?
Everybody tells you all your decisions are GREAT, even when they make absolutely no sense?
“Oh, you’re doing a 9-sided 3 wheels swerve drive again? Lookin’ great Robowranglers, as always!”
Side Note: For the life of me I cannot spell “Devastator” correctly the first time I type it. For some reason my brain defaults to “Devestator.”
By the way, people seem to be confused by the brick in the bottom of Devastator. This is “Mr. Cinderblock” - he has been a valuable and beloved member of the Robowranglers since 2016.
When we were tuning Renegade’s catapult, and I was WAY too grumpy about it… I started joking about: “I mean, it’s not too late to just throw a cinderblock on the drivetrain and go FULL DEFENSE this year.” (This system concept was named “Steve the Pirate” to go along with the other “Dodgeball” character names). When we were touring the construction progress of the new Robowrangler Shop we were introduced to Mr. Cinderblock. He’s been with us ever since!
For teams who want help focussing their efforts and building within their limits… I published a priority list for the 2019 game. Check it out. JVN Blog: JVN’s Priority List 2019.
At this point, I feel like 148 has a pretty good handle on our priority list. Our list is very similar to the published one, with a few rearranged items, mixed with our typical ridiculousness. (Our #1 priority is always: “Black Robots Go Faster” - since it’s a known fact.)
Time Studies & Scoring Benchmarks
We use “time trials” as form of prototyping and this very effectively cemented our priority list.
Using past robots, our past experiences with “acquisition” times, testing of typical “line up times” along with a healthy dose of imagination we were able to get some scoring cycle estimates.
We literally drive the old robots back and forth while measuring their cycle times. X019, Bolt, and Uppercut all made cameo appearances. We even used Uppercut’s elevator to simulate “lifting up” before scoring in the different positions.
“If you assume it takes you X seconds to pickup an object, Y seconds to score the object…” How fast can you fill the Cargo ship? How fast can you fill a Rocket? What happens if we put a defender robot in the way? How much time do you lose per “drop” or “missed alignment”? What % of the time do we expect to have drops? Factor in that percentage to get the average cycle calculation…
You can then use this data to do some basic “alliance scoring” analysis. What happens if all 3 robots score all match, and no one climbs? If we need to go back to the platform at 30 seconds to get on 254’s lift, how much did we lose in object-scoring? What is the % likelihood that our end-game plan fails - can we use buddy-climb stats from last year as a benchmark? At what % risk does it become worth doing?
Remember: Points vs. Time.
“How do we score as many points as possible? How do we score points as fast as possible?”
Our team has seen many of the comments “on the internet” about cycle times, including some people posting estimates about “what teams like 148” will be capable of. After reading these we’re 1- flattered, 2-very confident we are not “like 148” based on these expectations vs. our own data.
I hope more teams will take the time to do the analysis. All you need is an old robot, an open stretch of carpet and some tape to mark out the field. In my priority list I posted a few benchmarks but always hesitate to “step out there” with predictions based only on early data. I’ll say this: if you’re AVERAGING 6-8 per match, I think you’re going to be doing pretty darn good. (Note: FRC students frequently struggle to understand the difference between “best run ever” and “average”. (Note Note: this phenomenon does not only apply to FRC students.))
What do you think? (And more importantly, WHY do you think it?)
Prototyping & Field Construction
“How do you guys have time to do all this time based analysis?”
While all the time studies and prioritization stuff was happening… we also built a lot of stuff. Lots of prototype object manipulators. Lots of “plywood strapped onto old robots”. At one point X018 had an intake hanging off each side, and X019 was running cycles actual scoring cycles with a prototype disc manipulator. Lots of parallel paths happening on 148.
We Robowranglers do love our universal intakes!
This velcro conveyor belt was inspired by “Floppy” picks from the 1999 game.
Uppercut wanted to help out with this year’s game too! Uppercut likes spaaaaaaaace.
X018 saw X019 scoring and didn’t want to be left out!
While the rest of the team was hard at work, I was blogging and D-Row was doing… this:
“Give the people what they want!” I guess…
Packaging and System Design
The biggest challenge in this year’s game seems to be systems integration. We’ve got lots of prototypes that do amazing things. We’ve got a fairly good priority list. We have been struggling to fit “all that robot, into one robot.” Or at least struggling to do it with less than 15 individual actuations. (Bryan Culver taught me that individual actuations is a good way of quantifiably comparing “robot simplicity” - I posted a writeup about Robot Simplicity and how the historical Robowrangler robots compare. Check it out.)
We’re doing a ton of system concept work in CAD. Many people still question me when I say “the only prototyping I know how to do is in SOLIDWORKS, so I did a writeup showing some of the historical work we’ve done on the Robowranglers. JVN Blog: Prototyping in CAD. It’s amazing the things you can learn from a few colored blocks crammed into a robot sized space.
Robowrangler Student: “Is a scissor lift still an option?”
JVN: “Failure is ALWAYS an option.”
I created a set of “crude” field elements for crayola CAD. We isolated the parts we wanted in the Field Assembly, then saved them as a single part file. Makes for a lightweight way to see a design “on the real field” without loading up the big SOLIDWORKS assembly.
After doing some initial layouts and better understanding the packaging tradeoffs (things like: Wow, it really sucks how that outer ball roller is ALWAYS in the way of the disc pickup) the design team presented the larger team with a list of 15 questions. These questions have HUGE impact on the robot system.
Some of them we were able to answer using data from our time-trials. Others we were able to answer based on prototyping. Some have answers like: “Whatever the designers think will fit.” or “it’d be nice, if we can make it work.”
We still don’t have answers for all of them, but we’re getting closer.
Earlier this week we had the 2019 version of “Auto-palooza”.
This is an activity we do during Robowrangler Huddle in which we talk about strategies for autonomous mode. Wranglers will step up and say “I think we should do X, Y, and Z”. However - we’re not just talking about our robot’s actions, we role-play for entire hypothetical alliances.
“What happens if we’re partnered with 254 and playing against 118? What do we do, what does 254 do?”
It’s a fun activity, especially since students act out the actions of the robots.
However, since the Autonomous Period has been replaced by the Sandstorm Period, I guess it is "Sandstorm-palooza?
Unlike Autopalooza in 2018 this year we very quickly came up with the actions we planned to perform, we really struggled with the location for those actions. There are a LOT of goals on the field, and everyone seemed to have different opinions on which ones to use in the Sandstorm period.
Our decision? We decided NOT to decide. All the different strategic options involved very similar robot functions - so we said: “We’ll figure it out once we have the robot working and can do more simulation.
That’s one of the lessons I’ve learned as Lead Engineer. A few years ago, I would have tried to make a decision. Now - we’ll just figure it out later, no reason to decide now… we know what we need to know.
Perhaps the most important update… after a very exciting pre-season coming up to speed with the team, our newest Shop Dog made his 2019 debut!
This is Goddard.
His hooman is Garrison a new mentor on the Robowranglers (an alumni from FRC2468 and new VEX Robotics Engineer). We certainly APPRECIATE Garrison being around, but Goddard takes things to a new level. Unlike past Shop Dogs, Goddard makes appearances during Robowrangler Huddle - he walks around the inside of the circle saying “hello” to everyone and distracting them from whatever it is I’m saying. (Sometimes distracting ME from what I’m saying).
You can even see Goddard helping out during Sandstorm-palooza! (Well… trying to help out, he wasn’t very good at simulating robot actions.)
Is it possible to be 2 weeks behind, if we’re not even 2 weeks into the season yet? Just asking for a friend… This is Fine.
Are We Dumb?
I present to you the latest installment of “Are We Dumb?” without comment…