Stage 1: Research - Let’s talk strategy!
During the last 5 days, our strategy team has been busy with doing a lot of spreadsheets and calculations. Of course, we started with reading the manual and worked our way towards the solution for our week 1 regionals robot concept. Below, we’ve written down a quick overview of how the week went. Without further a do, here is the link to The Complete Spreadsheet for our Stage 1 Research phase.
Day 1: Saturday
We started all very excited about this year’s game Crescendo! Besides from being hyped, we were also scared by the amount of strategy this game brings. “Is the TRAP a trap?” 
Game Points Analysis
We’ve written down all points that can be scored in this years game and discussed the rules with each other to make sure that we all understand the game.
Detailed explanation
| Game Tasks | Auto | Teleop | RP | CP |
|---|---|---|---|---|
| Leave | 2 | |||
| Amp note | 2 | 1 | ||
| Speaker note (not amplified) | 5 | 2 | ||
| Speaker note (amplified | 5 | |||
| Park | 1 | |||
| Onstage (not spotlit) | 3 | |||
| Onstage (spotlit) | 4 | |||
| Harmony | 2 | |||
| Note in Trap | 5 | |||
| At least 18 (15 with coopertition) AMP or Speaker notes scored | 1 RP | |||
| At least 10 Stage points and 2 ONSTAGE robots | 1 RP | |||
| Tie | 1 RP | |||
| Win | 2 RP | |||
| Lose | 0 RP | |||
| Coopertition | 1 CP |
Game Definitions
This year’s field has a lot of different structures, some with names and some without. We added extra definitions to help us with explaining our thoughts to each other. As you might know, the Stage for example already has a name for each side: Right Stage, Left Stage and Center Stage. Although the naming is clear, it’s relative to both alliance walls instead of having the names relative to the field. For us, it works better to have all names field centric so that the naming is the same for both sides of the field (mirrored).
Detailed explanation
Robot Skills
We create a complete skills list. This list contains all possible robot skills that might be needed in the game. It is structured in such a way that everything a robot can do, can be split-up into these fundamental functions.
We created the skills list to explore all possibilities of the game and to make sure that we don’t miss a robot skill during the prototyping phase.
Detailed explanation
| Category | Robot Skills |
|---|---|
| Mobility | Drive |
| Drive under STAGE CORE | |
| Drive over NOTE | |
| Camera based driving | |
| Detach robot from chain | |
| Climb on SUBWOOFER | |
| Collect | Collect NOTE from floor |
| Collect horizontal NOTE against wall | |
| Collect vertical NOTE against wall | |
| Collect NOTE from SOURCE | |
| Collect from SUBWOOFER | |
| Control | Store single NOTE |
| Push single NOTE | |
| Constraint: NOTES may not get stuck | |
| Scoring | Score NOTE in AMP from AMP ZONE |
| Score NOTE in AMP from outside AMP ZONE | |
| Hold position while ONSTAGE | |
| Robot can get ONSTAGE on lowest point of chain | |
| Robot can get ONSTAGE on highest point of chain | |
| Robot can get ONSTAGE on any point of the chain | |
| Robot can traverse left/right ONSTAGE | |
| Robot can get ONSTAGE when another robot is already ON the same STAGE | |
| Score NOTE in TRAP | |
| Score NOTE in SPEAKER from SUBWOOFER front | |
| Score NOTE in SPEAKER from SUBWOOFER side | |
| Score NOTE in SPEAKER from PODIUM | |
| Score NOTE in SPEAKER from WING | |
| Score NOTE in SPEAKER from outside WINGS | |
| Eject NOTE on ground | |
| Eject NOTE with height | |
| Lift 1 additional robot ONSTAGE | |
| Lift 2 additional robot ONSTAGE | |
| Attach to ONSTAGE alliance partner | |
| Alignment | Align robot with AMP |
| Align robot with SOURCE | |
| Align robot with PODIUM | |
| Align robot with TRAP | |
| Align robot with chain | |
| Align robot with SUBWOOFER front | |
| Align robot with SUBWOOFER side | |
| Align robot with NOTE |
Day 2: Sunday
We’ve made a lot of progress on Sunday! We took a deep dive into auto/teleop cycles, analyzed Stage points and looked at old games that have similar game elements (2013/2017/2023).
Auto Cycles
We generated all kinds of auto paths that our possible for this year’s game. We then eliminated the inefficient paths and thought about what paths we can combine to make good auto routines.
Detailed explanation
| Start position | Collect position | Score position |
|---|---|---|
| subwoofer 1 | free note 1 | podium |
| subwoofer 1 | contested note 1 | wing 1 |
| subwoofer 1 | contested note 2 | wing 1 |
| subwoofer 1 | contested note 2 | wing 2 |
| subwoofer front | contested note 4 | wing 2 |
| subwoofer 2 | free note 3 | wing 2 |
| subwoofer 2 | contested note 4 | wing 2 |
| subwoofer 2 | contested note 5 | wing 2 |
| subwoofer front | contested note 5 | wing 2 |
| podium | free note 2 | wing 2 |
| podium | free note 3 | wing 2 |
| podium | contested note 1 | wing 1 |
| podium | contested note 2 | wing 1 |
| podium | contested note 2 | wing 2 |
| podium | contested note 3 | wing 2 |
| podium | contested note 4 | wing 2 |
| wing 1 | free note 1 | podium |
| wing 1 | contested note 1 | wing 1 |
| wing 1 | contested note 2 | wing 1 |
| wing 1 | contested note 2 | wing 2 |
| wing 1 | contested note 3 | wing 2 |
| wing 1 | contested note 4 | wing 2 |
| wing 1 | x | neutral zone |
| wing 2 | free note 3 | wing 2 |
| wing 2 | free note 2 | wing 2 |
| wing 2 | contested note 5 | wing 2 |
| wing 2 | contested note 4 | wing 2 |
| subwoofer front | free note 2 | wing 2 |
| subwoofer front | free note 3 | wing 2 |
| subwoofer 1 | x | neutral zone |
| subwoofer 2 | x | neutral zone |
| wing 2 | contested note 3 | wing 2 |
Teleop Cycles
For the teleop cycles, we basically did the same as for auto cycles. We collected all possible combinations and filtered on the most relevant ones.
Detailed explanation
| Start position | First Wingline | Second Wingline | End position |
|---|---|---|---|
| Subwoofer | 1 | 1 | Source |
| Subwoofer | 2 | 1 | Source |
| Subwoofer | Stage | 1 | Source |
| Subwoofer | 2 | Stage | Source |
| Subwoofer | 1 | Stage | Source |
| Amp | 1 | 1 | Source |
| Amp | 2 | 1 | Source |
| Amp | Stage | 1 | Source |
| Amp | 2 | Stage | Source |
| Subwoofer | 1 | Contest note 1 | |
| Subwoofer | 2 | Contest note 1 | |
| Subwoofer | Stage | Contest note 1 | |
| Subwoofer | 1 | Contest note 2 | |
| Subwoofer | 2 | Contest note 2 | |
| Subwoofer | Stage | Contest note 2 | |
| Subwoofer | 1 | Contest note 3 | |
| Subwoofer | 2 | Contest note 3 | |
| Subwoofer | Stage | Contest note 3 | |
| Subwoofer | 2 | Contest note 4 | |
| Subwoofer | Stage | Contest note 4 | |
| Subwoofer | 2 | Contest note 5 | |
| Amp | 1 | Contest note 1 | |
| Amp | 2 | Contest note 1 | |
| Amp | Stage | Contest note 1 | |
| Amp | 2 | Contest note 2 | |
| Amp | Stage | Contest note 2 | |
| Amp | 2 | Contest note 3 | |
| Amp | Stage | Contest note 3 | |
| Amp | 2 | Contest note 4 | |
| Amp | 2 | Contest note 5 | |
| Source | 1 | Stage | Stage 1 |
| Source | 1 | Stage | Stage center |
| Source | 1 | Stage | Stage 2 |
| Subwoofer | Stage | Stage 1 | |
| Subwoofer | Stage | Stage center | |
| Subwoofer | Stage | Stage 2 | |
| Amp | Stage | Stage 1 | |
| Amp | Stage | Stage center | |
| Amp | Stage | Stage 2 | |
| Source | 1 | 1 | Subwoofer |
| Source | 1 | 2 | Subwoofer |
| Source | 1 | Stage | Subwoofer |
| Source | Stage | 2 | Subwoofer |
| Source | 1 | 1 | Amp |
| Source | 1 | 2 | Amp |
| Source | 1 | Stage | Amp |
| Source | Stage | 2 | Amp |
| Contest note 1 | 1 | Subwoofer | |
| Contest note 1 | 2 | Subwoofer | |
| Contest note 1 | Stage | Subwoofer | |
| Contest note 2 | 2 | Subwoofer | |
| Contest note 2 | Stage | Subwoofer | |
| Contest note 3 | 2 | Subwoofer | |
| Contest note 3 | Stage | Subwoofer | |
| Contest note 4 | 2 | Subwoofer | |
| Contest note 5 | 2 | Subwoofer | |
| Contest note 1 | 1 | Amp | |
| Contest note 1 | 2 | Amp | |
| Contest note 1 | Stage | Amp | |
| Contest note 2 | 2 | Amp | |
| Contest note 2 | Stage | Amp | |
| Contest note 3 | 2 | Amp | |
| Contest note 3 | Stage | Amp | |
| Contest note 4 | 2 | Amp | |
| Contest note 5 | 2 | Amp |
Stage Analysis
We analyzed the stage to see what combinations can get us closest to the Ensemble RP.
Detailed explanation
| Total options to reach 10 points | 126 |
|---|---|
| No Spotlit needed | 40 |
| No Spotlit or Harmony needed | 18 |
| Only 2 robots needed | 57 |
| All 3 robots needed | 69 |
| No Trap needed | 14 |
Old Games
We looked into 2013/2017/2023 to estimate cycle times for this season.
Detailed explanation
This analysis results in an expected cycle time of about 10 seconds for the best robots in the world. To calculate this, we had to make some assumptions. For example, we compensated for drivetrain speed by dividing them by 2. This is because tank drivetrains in 2013 are about twice as slow as modern swerve drivetrains in 2024.
Referenced matches:
2013 Einstein match
https://youtu.be/fnVXMZNi1nk?si=fx9hKoNsntVqVAfk
2017 Einstein match
https://youtu.be/dBcAwU8I48E?si=17Q5kzAB_c8pIv5X
2023 Einstein match
https://youtu.be/PFW2Sc3shaA?si=QzrsLk_LfAInUYRC
Day 3: Monday
Goal of day 3 was to get a deeper understanding of how the game is going to be played and what the meta will be at different stages in the season.
Ranking Points
We did some further analysis on both ranking points.
Detailed explanation
Melody RP
For the Melody RP, we used our estimated cycle time analyses from day 2 and tried to convert those into good estimations on how many notes we can score on our own.
We estimated that we can score 14 Notes solo, meaning that getting the RP on our own is going to be hard but still possible with a lower threshold from using the Coopertition button.
Ensemble RP
This Ranking Point is just theoretically inpossible to score on your own (except for robots that have a buddy climb system). Scoring in the trap makes it easier to score the RP, but you would still need a robot to climb. High note scoring by the human player also makes sure that 2 robots is enough, but they have to hang from the same chain.
| Spotlit | Not spotlit | |
|---|---|---|
| 2 robots can climb | Same chain | Not possible without trap |
| 3 robots can climb | Different chains | 2 same chain, 1 other |
Human Player
We talked about the importance of the human player(s) this year.
Detailed explanation
We concluded the following points:
Human Player AMP
- Click Coopertition button
- Wait with clicking the bonus button till first shot hits
- Coordinate with drivecoaches when to get teams ready to shoot
- Throw rings onto the mic or introduce them close to a robot ready to shoot
Human Player Source
- Train with signing to get ready
- Train with dropping the notes just on time for the robots to pick it up
- Coordinate with drive coaches to always be ready with a ring
Match Strategies
We’ve written down possible alliance matchups and strategies they can play. For example, double Offense + one Defense or 2 robots SPEAKER + 1 robot AMP
Detailed explanation
Autonomous Strategies
- 3 robots score preloaded in speaker or amp
- 2 robots score preloaded in speaker and 1 scores preloaded in amp
- 1 robot scores preloaded in speaker and 2 score preloaded in amp
- 3 robots score preloaded. 2 will go contested notes + 1 will go free notes
- 3 robots score preloaded. 1 will go contested notes + 2 will go free notes
- 3 robots score preloaded. 3 will go contested notes
- 3 robots score preloaded. 1 will go free notes and scores in speaker + 2 will go contested notes and collect or score speaker
- 3 robots score preloaded. 2 will go free notes and score in speaker + 1 will go contested (collects them and shoots them to own wing)
- 3 robots score preloaded. Subwoofer front will go free notes and scores them in speaker. Subwoofer 1 and 2 will go contested. 1 will collect and score in speaker + 2 will collect and shoot to wing (can also be the other way around)
Teleop Strategies
- 3 robots do speaker/amp cycles
- 2 robots do speaker cycles + 1 does defense
- 1 robot does speaker cycles + 2 do defense
- 2 robots do speaker cycles + 1 robot feeds notes
- 1 robot does speaker cycles + 2 robots feed notes
- 1 robot does speaker cycles + 1 robot does defense + 1 robot feeds notes
- 1 robot does speaker cycles + 1 robot feeds + 1 robot focusses on getting the speaker amplified (amp cycles) (when amplified 2 or 3 robots score speaker with notes that have been fed)
- 2 robots do speaker cycles + 1 robot focusses on getting the speaker amplified (amp cycles) (when amplified 2 or 3 robots score speaker)
- 2 robots do amp cycles + 1 robot feeds notes
- 1 robot does amp cycles + 2 do defense
- 2 robots do amp cycles + 1 does defense
- 1 robot does amp cycles + 2 robots feed notes
- 1 robot does amp cycles + 1 robot does defense + 1 robot feeds notes
Cycle Times
We have tried teleop cycles with our SDS swerve drivetrain and computed the averages.
Detailed explanation
| Starting position | End position | Average time |
|---|---|---|
| Subwoofer | Contested notes | 3 seconds |
| Subwoofer | Source | 5 seconds |
| Amp | Source | 5 seconds |
| Source | Wing | 4 seconds |
Day 4: Tuesday
Our final day of actively collecting and computing data. We used all our aquired knowledge to think about trade-offs and come up with robot concepts.
Trade-offs
There are a lot of trade-offs in this game. We summed up all the ups and downs for each trade-off.
Detailed explanation
Trap
Positives:
- Helps with reaching the Ensemble RP
- A few more points
Negatives:
- Resource expensive
- Doesn’t add a lot of value in playoffs
Amp
Positives:
- Boosts points by a lot
- Ability to help alliance partner score more points
- Able to use the coopertition button
- Helps with reaching the Melody RP
Negatives:
- Can be resource expensive
- Not worth many points without decent alliance partners
Speaker
Positives:
- A lot of points
- Helps with reaching the Melody RP
Negatives:
- Can be resource expensive
X% Rule
The X% rule says something about how many points we should score at our week 1 regional, to have a high probability of winning the event.
Detailed explanation
We first determined what an Einstein performing team would score (that is 100%). Based on that we’ve taken X% of those points so we know the amount of points we have to score. The X variable is determined by where and when we play our regionals. For California week 1, we would need 70% of the points the best robots in the world score on their own.
100% Robot Concepts
We determined our 100% by looking at our cycle analyses.
Detailed explanation
Based on what we tested and on our cycle analyses from previous years, we think an average cycle time of 10 seconds is realistic for an Einstein performing robot. After that it is just a simple puzzle of filling in what cycle is done how often. Eventually we landed on 72 points. This value should be equal to the OPR of the best robot in the world if the estimation is 100% correct. Since we can’t look into the future, this value might be a bit lower/higher, but we feel like 72 points is a realistic estimation.
70% Robot Concepts
Now that we have the 100% robot scores, we can simply take 70% of those points to get our week 1 target.
Detailed explanation
70% of 72 is about 50 points. This means we need to have a robot that is able to score at least ~50 points to have a high probability of winning the event.
We adjusted our cycle time to reach 50 points with our concepts and this results in a cycle time of about 12 seconds. These 50 points can be reached by multiple different cycle strategies and different robots.
Day 5: Wednesday
This is the day of the final decision which kind of robot TR will pursuit this season.
Potential Robot Concepts
Before the meeting we had with the rest of the team, we wrote down 5 interesting robot concepts that all have their ups and downs.
Detailed explanation
These potential robot concepts are the following:
| Ground intake | Climber | Speaker | Amp | Trap | |
|---|---|---|---|---|---|
| 1 |  |
|
|
|
 |
| 2 | |
|
|
|
|
| 3 | |
|
|
|
|
| 4 | |
|
|
|
|
| 5 | |
|
|
|
|
Final Decision for Robot Skills
We considered all concepts. Their scoring potential, RP potential and required resources.
Detailed explanation
We choose for the Amp/Speaker robot without Trap mechanism. The reasons for this are mostly because we don’t want to take too much risk by making a robot that can do everything in the game. On the other side, we also still wanted to score enough points to reach the 70% / 50 points we need to perform well at our week 1 regional in California.
How the robot will be designed still depends on the archetype. We’re considering multiple options regarding the design, but the decision for doing Speaker/Amp only holds. We’ll post more about archetypes later this week!
Reflecting on the last few days
I’m proud to say that we’ve completed all of our tasks before the deadlines. I really enjoyed sharing our knowledge with you all. Your positive reactions really boosted our enthousiasm for analysing this game! Thank you very much and let’s keep on making progress together 
Written by:
@Bjorn - Lead Data Driven Decision Making