I’ve noticed that almost no one goes for the colour wheel this year because it was deemed as not worth it. How do people decide what is/isn’t worth it for every year? Our team felt like prioritizing was an area to improve on so I thought this would be a good place to learn from. Thanks for any answers.
Bit older - https://blog.thebluealliance.com/2014/10/10/sa-strategy-by-numbers/ (and it seems that the transition onto TBA Blog broke some stuff)
https://blog.thebluealliance.com/2014/10/10/sa-dozer-does-it/ shows that doing something consistent is often better than most teams.
In the case of the wheel of fortune, my exact thought process was how many points is it worth with various circumstances.
0 balls - 0 points
10 balls - 0 points
Basically it showed unless I could be 100% sure those balls got scored there was no incentive to fuss with it. This moved it down my priority list below scoring balls.
Disclaimer - I wasn’t involved in FRC this year very much having personal life things get in the way.
On kickoff day, I generally try to starting thinking about tasks in terms of two questions: how much is it worth, and how hard is it to complete? The Control Panel, while not especially hard to build a mechanism for, is worth at most 10 points until 49 balls are scored. In other words, if you can’t hit 49 balls in a match, your control panel mechanism is worth 10 points, or five outer port balls, which isn’t a ton of points, and a team needs to be able to score power cells anyway to be competitive.
Deciding all comes down to strategy. Normally, we like to look at all the different ways to play the game. Then, decide how many cycles teams can do in a match and find what strategies earn the most points. You can then find areas are most important.
Everything looks better
Be careful when looking at strategy to not fall into the common pitfall of “doing everything is the best strategy”. Very few teams are well equipped to do everything. Understanding how your team is can handle situations is very important.
Imagine you have 10 credits to build a robot. Do you want to do three things 3/10 or one thing 10/10? Figure out the best thing to do, and then do it really really well.
There is another level I’d throw in here. There are multiple levels of play. There are qualification matches. Regional Elims. Division Championship matches. Worlds Qualification matches. And Einstein.
First ask where am I shooting for (in reality) this year
In my first team in their first year, we were were shooting for middle of the field. Two years later we were shooting for a must have 2nd pick for a regional elim alliance. This year my team was shooting for a top 3 ranking in a worlds division.
Now that you’ve got your goal in mind, ask yourself, what parts of play are going to get my team to that goal. That’s what your priority is.
This year, we felt that 50% of matches in Quals at worlds would need a color wheel robot. We didn’t push hard for it for our first regional. We had a backup (low tech) plan that might have worked if it was needed. We were actively working on it for our second regional and would have had a good device. We were going to streamline and make it super clean for worlds.
So, make your plan. Build against your plan to be the best it can be.
PS: It took 2 weeks of build season for our game theory people to settle down on realistic numbers for the color wheel. So we had to modify our initial plan a little as we went along.
For us, it started with a simple goal: Build a robot which scores the most points per second.
With this goal in mind, we analyzed the rules book, and started to make some reasonable guesses as to what score rates different robots would have. We ended up with roughly three levels of “Cheesy Poofs”, “Totally Average”, and “Bucket of Bolts”.
From this, we were able to quickly figure out that:
- Climbing is worth a metric ton of points, which can be scored in a relatively small quantity of time.
- Shooting upper goal 2’s and 3’s is the next most points-rich score methodology
- Picking up power cells off the ground was going to be critical, especially considering flood possibilities. A solid ground intake was next on the list.
- Color wheel appeared necessary to progress all the way through the ranking points, but we were willing to bet on the fact that someone else on the alliance could do it. Even still, we had enough time to add something to the robot to accomplish it anyway. But it got almost zero attention till week 5.
We independently decided that dumping in the lower goal will not be literally twice as fast as upper goal shots, so we shouldn’t pursue it. That being said, we were confident in our teams ability to construct some sort of shooter.
Really, the “how” is flexible every year, since the game changes. The common keys to success, IMO, include:
- Careful reading and analysis of the game manual.
- Honest estimation of scoring abilities of your own team, and others
- Knowing your own team’s restrictions and building within them.
For better or worse, doing these things well requires lots of experience. Research will definitely help, but there’s not much of a substitute for senior students or mentors who have the history to know what is and isn’t reasonable.
From there, it’s a bunch of random chance - did your guesses turn out to be correct?
A lot teams usually have a couple really in-depth strategy meetings early on in the season to determine priorities. This includes but is not limited to, reading and re-reading the rules over and over again. For our strategy meeting, we estimated cycle times as well (cycle times for this game would be the amount of time it takes to grab 5 balls and go shoot them).
After we did some very questionable math, we determined that unless you have at LEAST two amazing robots doing reasonably successful cycles, finishing stage three of the color wheel wasn’t possible. Our decision was made by factoring in defense, potential driver errors, and potential jamming issues with the power cells.
For determining other priorities, we (1912) usually do an “efficiency” estimate. We basically determine what function (climbing, color wheel, shooting PC) gives us the most points for the least amount of time. This year, climbing not only gave you the most points in the least amount of time, but also gave the chance to help out your teammates with an RP.
I would just add that I saw a fair number of robots at Utah that had color panel mechanisms, but just didn’t use them because not enough balls were being scored. Just because it wasn’t used in Weeks 1 or 2, doesn’t imply everyone sat down on Kickoff Day and decided it wasn’t worth building.
My team decided we would focus on only doing low goal and control panel, with the goal of (by the end of the season) being capable of getting that rank point on our own. It was a stretch goal for sure, but given that we had Week 2 matches where we (on our own, not our alliance) scored over 20 balls, it’s not a crazy stretch to think that we could have gotten there by the end of our Week 5 event, and pretty reasonable if we had alliance partners that could score a decent number of balls. We even had one match towards the end of Utah where we could have done the first stage of the control panel, if our drive coach had noticed.
As a final note, we chose the objectives we did in part because we had an extremely young team this year (2:1 rookie:returning ratio), and haven’t successfully built a shooter or climber in the past handful of years. I still maintain this was a good strategy for us this year, but not necessarily a good strategy for other teams with different capabilities.
For our team, when the season got released our main goal is to figure out what we wanted to excel at. Our team mainly likes to focus on one thing in the game, and do it well. We usually take votes and discuss everyone’s thoughts and opinions. Eventually, we brig it down to a few things we would like to do. Once we get to this point, we prototype our ideas, and figure out what works and what doesn’t. This may be a little tough sometimes, because you always need to have the knowledge of the game in mind, and also think about strategy all of the time.
One thing we do to help ourselves anticipate how the game will be played is run simulated matches. We start by deciding the capabilities we want to give to each robot, typically we put a top-tier, average, and basic robot on each alliance. We then run through the match 15-30 seconds at a time, with a small group controlling each robot.
Only a few teams ever spun it. Many more teams had a mechanism for it and never even came close to scoring enough balls to mess with it. All of those other teams wasted their time, energy, and weight.
When tasks are conditional based on game conditions, you need to honestly evaluate how likely those conditions are going to occur. This varies year to year based on game piece availability, cycle time, etc. but it should have been clear to many that 49 balls was not going to happen except at the very highest levels of play - and if you’re on an alliance with 49 balls scored, you probably have an elite partner who can do it anyway. If not, just do one more cycle.
Also, don’t discount the idea of doing a few Organic Robot matches (i.e. students pretending to be robots walking around the field and pretending to play the game). Once you’ve talked about things like expected cycle times, how many balls per cycle, how long to get lined up and climb, etc. Have a 3 on 3 alliance of students walk around the field (or a taped out area in the parking lot simulating a field) and run through a couple match scenarios. It feels dumb, but it makes everything a lot more tangible, and you’ll notice things that you didn’t think of.
Cycles per match. If you can guarantee X balls scored vs Y wheel spins and the points for the balls is greater, there is no point in spinning the wheel.
We discuss some of our prioritization techniques here:
We’ve seen that the top teams are able to do everything in a match (and do it well).
We’ve seen that teams that specialize in a single task are usually not chosen during alliance selection (there are exceptions).
We needed to be able to do every task.
The team prioritized our build.
- Pickup and shooting
- Control Panel (one mentor suggested that this would not be a high priority for week 1 regionals)
Unfortunately, we spent far too much time on the first to devote enough resources on designing and building a climber or control panel manipulator.
By Palmetto, we had a successful pickup/shooter. Our climber was great; especially if we were the only robot on the field . Although the basic design of our hook delivery system was great (we saw many teams use similar devices), it was susceptible to failure (understatement).
First thing to do is read the rules and find out how to score points and how to get ranking points.
Also consider what mechanisms are needed. It was clear that the easier bonus RP was climbing and that points from climbing were worth a lot more than several cycles scoring PCs. Also elevators have been in the last two games so finding good examples should be easy. Climbing would be something that would be a requirement for a playoff team.
So a top team would need:
- PC Hopper/Index
I’ll go over each one.
Drivetrain. Use a drivetrain you know how to build quickly and one that you know is reliable.
Intake. Either ground pickup or loading zone pickup. Looking at the game you can see that ground pickup offers more options. Loading zone pickup is limited by how well your opponents score. There are two types of ground pickup, through bumper or over bumper. Through bumper is not affected by defense but at it’s max is 12" less then max over bumper. Over bumper would need an additional motor or pneumatic. Through bumper will affect PC hopper/indexing more than over bumper. Through bumper needs better driver skill. Both through bumper and over bumper intakes have examples of mechanisms from past games so you can copy the best.
Shooter. Again, there are several excellent examples from previous years games to copy whether it’s a hooded shooter or a dual wheel shooter.
Climber. 2018 has great examples of climbers, both single point and dual point climbers. Single climbers are smaller to package but you need to be concerned with violating G18 (extending 12" outside frame perimeter). Dual point climbers take more space but are easier to solo balance, shouldn’t violate G18, can still work if only one point grabs switch.
Spinner. Something new.
PC Hopper/index. Using space available and referencing examples from previous games…
The next part of the strategy is deciding on whether you want to be a tall or short bot.
Tall robot. Easier climb mechanism. Less likely for shot to be blocked. Has to either shoot from behind control panel OR has to go through rendezvous zone to score.
Low robot. Tougher climb mechanism. Shot easier to be blocked. Protected trench run.
If you are building a shooter, how are you going to aim? Driver control? Robot aiming? Turret?
Where are you planning on scoring from (Target zone, Initiation line, End of trench, behind control panel)?
How does defense affect where you are planning on scoring from?
All of these affect the type of mechanisms you choose and their complexity you are going to end up with.
I’d say its different every year. Usually we focus on what can get us the most points, while still keeping in mind that we are a high school robotics team and there are limits to what we can do in a certain amount of time.
As a team that has existed for a while we usually tend to focus the elements of game that are more unique and will score the most points. Most games you encounter though your years of FIRST will have some object that you pick and place or shoot. So making mechanisms like this have become second nature to us.
For example, last year we knew we had to pick up panels and cargo and place them. Because we knew that this would be fairly easy to do, we put it on the back burner and decided that it was most important that we were able to climb very quickly. With this in mind we designed our entire robot based around our climber and then buit the rest.
Its honestly different for most teams and games, but what you want to focus on is what will get you the most points for what you believe your team is able to do in a certain amount of time.
What do you mean when yoy say that through bumper pickup isn’t affected by defense?
An over bumper intake goes outside the frame perimeter and can be damaged by defense or running into a non-movable field element. It’s a risk that has to be considered when comparing the two mechanisms. Through bumper intakes are more protected. Generally you’ll find that over the bumper intakes are quicker because the driver doesn’t have to be as precise, plus you can use materials that are more flexible and can withstand some contact.
Oh, ok. Thanks for clarifying.