Team Rembrandts 2023 building season update 3
Game Analysis
Having a successful season requires building a strong robot. Building a strong robot requires a strong strategy. Having a strong strategy requires a high-level understanding of the game. This year’s game, FIRST Charge up presented by Gene Haas Foundation poses a new challenge to teams once again.
This analysis is a follow-up to the game analysis we already started in our Kickoff / Reaction Post. We’ve started analysis on several different things, chief among them: Where do we want to score? and how quickly could we score? Both are going to be invaluable in not only deciding what we should build but also how to test them, when are they performing to the level we need them to perform?
Location Scoring
To start by stating the obvious: scoring links is going to be very beneficial. An easy calculation shows that even when scored on the highest level it gives you a 33% scoring bonus over three unlinked cones or cubes. This does mean that it seems necessary to either score on the low level a lot or to be able to score both cubes and cones.
We’ve narrowed the options down to three ways that we think we could do and are worth looking into. These are: scoring everywhere with both cubes and cones, scoring on the middle and low levels with both cubes and cones, and scoring on the high level and lower with cubes, but on the middle and lower level with cones.
As a short reminder, this is how that would look points-wise, with on the y-axis how much cargo was scored in autonomous and on the x-axis how much cargo was scored in tele-op. For these charts, we assume that you always make as many links as possible.
Scoring everywhere with both cubes and cones:
Scoring on the middle and lower levels with both cubes and cones:
Scoring on the high level and lower with cubes but on the middle and lower with cones:
As expected, scoring high and lower is obviously scoring a lot more points, but the third option is a lot closer in certain scenarios when you’re not scoring as much. The reason this third option was chosen is, because we believe that we could potentially shoot cubes out of our intake, this would make it an easier mechanism for the cones and would also be slightly faster when scoring them, which would decrease our cycle time.
When you compare the high and lower scoring option with only scoring cubes high you do see that after a couple of cycles, you need to start scoring way more cycles to keep up. In the chart below you can see how many tele-op cycles extra you would need to outscore it this way.
After you score 2 or more cycles in an entire match it becomes very tough to outscore high everything, because you can’t link the cubes on the highest rungs so they become basically bad decisions to score unless you can’t reach the 3 more pieces of cargo to get a link anyways. Not only would this be very hard to predict with how chaotic matches can be, but even if you could, with 3 or more cycles certainly being realistic it becomes very unlikely that we could score 2 cycles more just because shooting cubes would be slightly faster (especially with full court cycles this year) or because a less development time to just score middle would give us more practice time.
Cycle time analysis
Field layout
How are we going to make a strong cycle time analysis without understanding the field first? Get it, FIRST . . . uhm yeah Right, Back to the topic.
The field for Charge Up has quite some interesting aspects to share.
First, we’ll look at the layout of the field and check which zones are the most relevant places on the field to be. You can see it as some sort of “health map”. The zones indicated by a number are the “places to be”. Not only does this give an indication of driving distances but it also allows for a very first strategic discussion.
For example, where on the field would you want to play defense, and is it worth it? Quite a big question to ask yourself.
Of course, each zone on the field has its own specific function in the game. In the table below we share what these functions are and how they affect gameplay. In the comments, we’ll note important limits or factors to keep in mind.
At this point, we have not yet looked into scoring data or game rules. These follow as we understand how the field works.
Scoring table and time assumptions
Let’s look at the scoring table. FIRST has made it a good tradition to add this to the game manual! We’ll only have to look it up and use it for our analysis.
In this game, one of the first aspects to notice is that scoring on different levels actually does give different point valuations. You might remember 2019, where the scoring position didn’t mean anything to the number of points an alliance would gain. Only the scored game piece mattered. In this game, it’s the other way around. Another aspect that immediately caught our eyes is the importance of strategic scoring. The scoring location in this game can impact your score and your ranking due to the link and co-opertition system.
It really boils down to this: It doesn’t matter what you score, as long as you place your game piece in the right spot. Now, this isn’t completely true as you can’t score cones or cubes on every scoring location. Some are only for one type of game piece but in essence, this is how we look at it.
In order to estimate times we’ll need to do some assumptions based on previous years. We’ll go through these, one by one. As you might expect, changing these assumptions will have an impact on your cycle time which in turn impacts a robots overall performance.
Autonomous cycles
The pieces for our analysis are set in place. Let’s be honest, this is the part you’re reading this blog for, aren’t you? Before we do, we’ll have to keep in mind our starting position. This can be any spot within the community. We’ll assume 3 starting positions, but to keep this post slightly brief we are only focusing on one start position, the other two will look very similar anyways. Now the analysis. We’ll use the scoring table and assumption table to give ourselves an indication for each cycle.
The main question here is simple: What do we think is possible within 15 seconds?
2 game pieces high → Docking + Engaging (Position 1)
领英推荐
3 Game piece and no Docking + Engaging is 5,80 sec.
Without engaging the score becomes: 15.
Drive time*1,5 because driving over the charging station takes more time.
3 game pieces high → Docking + Engaging (Position 1)
3 Game piece and no Docking + Engaging is 11,65 sec.
Without engaging the score becomes: 15.
Drive time*1,5 because driving over the charging station takes more time.
Tele-operated cycles
For tele-op cycles, we pretty much want to find out how quickly we can drive back and forth on the field. This game has a full field cycle. We haven’t calculated cycle times for the staging mark game pieces. The reason for this is because these aren’t consistent pick-ups as many things can happen during the game that influences their position. Besides this, some of these might be placed already once tele- op starts.
We can find cycle time by understanding pick and place times and calculating how long the robot travels from point A to point B and back to point A. Now the pick-up and place times depend on your subsystems. These estimates might change depending on proto-type results. The travel time can be found if we know the distance and robot travel speed in m/s.
The first figure is a rough straight path for the robot to travel. We purposefully went for these points as their coordinates can be found in the official field drawings. In order to keep the calculations simple we’ve not determined individual times for each scoring position. At this point that would be overdoing things due to the number of assumptions in our variables.
As you can see, the distance can be calculated by using geometry. The right top corner of the field in the image is taken as a reference, you can see the yellow dot. Now you might say, well, but what about the bottom-scoring position? You do have to drive quite a bit further right? Wouldn’t that take longer?
The answer is: yes it does. The difference is about 1 meter of drive distance. You can pretty much fill in the blanks. The result is about 1 to 1,5 seconds extra per cycle.
Assuming a scoring position as starting point. Then moving to the collect position, then moving back to the scoring position.
So: 3->2->1->2->3
Lessons Learned
With our analysis, we can conclude several things. First, let’s look at our autonomous cycle times.
You spin me right 'round, baby. Right 'round like a record, baby
Spindexer
As the team was working over the past week, we discussed many times how a robot from last year could play this year’s game scoring low and likely mid with a shooter for cubes. This discussion led to rewatching reveal videos from 2020-2022, which lead to the idea…
What if we could use a spindexer… for CONES? So the team set to work, and here are the results:
First concept
Fixed Latch
Flexible Latch
Next Steps
The next steps are to move our concepts into SolidWorks 3D CAD and see if we can make it all fit together. This is an age-old problem in FIRST robotics, it’s systems engineering time! For 4481 this challenge takes a whole new level. If you followed along with us last season you know that we try to follow a “unit build philosophy” that allows us to build modular subsystems that can be disassembled into check-on bag size sections so that we can fly with our robot and reassemble it when we arrive for competition.
This building season update was written by: Gijs de Veer , Tijn van Bommel , and Justin Foss .