In-depth Strategy Analysis for your perusal - Team 2374

[SirTasty]

Introduction
Greetings, Chief Delphi!

I wrote this paper for the benefit of my team, but I think there are some insights here that I would like to share with the rest of the CD community, so that I can get feedback and maybe help improve another team's robot. Essentially it outlines the results of our initial brainstorming process: First, it lists and supports the insights we made into the nature of the game, then examines the priorities we decided on based on our beliefs about the game will be played, and finally it explains the design decisions we have made to date and how they further our priorities.

I am a senior, and currently the systems engineering and design lead for team 2374, the CrusaderBots of Jesuit High School, in Portland, OR. This is just our fourth year in the business and we've still got a lot to learn, but I feel like, after four years of watching how the strategy works in this competition, I have a good handle on what makes a good robot. I'd like to hear from some more experienced team members (maybe some long-time mentors?)about how their experience does/does not fall in line with my points here.

Sorry for the wall of text, but there is no tl;dr here. I think reading the whole thing will be worth your time. Cheers and good luck!

Underlying Principles
Immediately after kickoff, we agreed on some basic premises around which to build our strategy:

1. Focusing on defense is an INFERIOR strategy. I’m surprised at the number of people who favor defense in the “offense vs. defense” threads on Chief Delphi. Historically, scores at Einstein are always very high, and usually winning alliances at regionals have three strong offensive robots; strong defensive robots are not often chosen, unless they bring some truly unique quality to the table-for instance, during overdrive we were selected for an alliance that went to the finals and had the ability to remove balls from the overpass, which allowed us to swing the game with some clutch late-game plays. This year, the penalties for encroachment on the alliance zone and lanes will be substantial, and the pinning rules are very strict, and defensive teams can count on being penalized for these regularly. The ranking rules this year, like last year, have also been designed to be unfriendly to defensive teams: while WL record is the primary determiner of ranking, FIRST has insisted on keeping approximately the same ranking points system to break ties. Last year we had the best defensive robot at our regional, and the more we won, the more we went down in the rankings. The rules have been tweaked, and I expect them to be a little bit less silly this year, but being good at defense will still not help a team’s ranking nearly as much as being good at offense. Lastly, defense is easy. Get in another robot’s way, push it around, pin them to walls, block them from getting into/out of lane, etc. If we fail at offense, we can still play defense and do well.

2. One minibot will not perform significantly better or worse than another. Just like every other team in the world, we had our own stored-energy minibot rocket plans until that pesky <G19> update. Now that that’s gone, we are all limited to the same components, and there’s not a lot of room left for innovation. Put on a good motor, make it as light as possible, and there’s not much else you can do to improve that bot. Because of this:

3. Teams have the greatest opportunity to affect the outcome of the match if they focus on hanging game pieces. If you can’t significantly improve your odds by focusing on your minibot, it’s obviously better to concentrate on something over which you have more influence. Notice that I didn’t say the minibots won’t decide the game; they may. However, you won’t get a lot of mileage out of focusing on them.

4. The Logo Piece game will be won or lost in the top row. Perhaps this is an overstatement, I think the middle row will also have an impact on the game, but the bottom row is not significant. The points scale majorly as the logos move to higher rows. A logo piece on the bottom row is worth only one point; an ubertube on that row only increases that score by one, for a total of two. That same configuration on the top row is worth six points – four more! Add in the fact that a completed logo doubles the value of the row, and you can see that gap widen even more. Three logos and one ubertube in the top row is worth 24 points – in the bottom row, it’s only worth eight.

5. Rapid DEPLOYMENT of the MiniBot is a critical factor in winning the minibot race. This isn’t rocket science. If everyone is the same speed, the one with the head start will win.

Priorities
1. Efficient game piece placement

a. Mechanically easy delivery

b. Ability to deliver reliably in the top row

c. Effective use of the sensor suite for placement

2. Efficient game piece pickup/control

a. From feeder station

b. From ground (optional)

3. Driving power/traction
4. Fast/efficient minibot deployment

a. Effective use of sensor suite to get the robot into deployment position.

b. Mechanically simple minibot delivery system

5. Speed
6. Maneuverability

a. Ability to move into position to score game pieces

b. Ability to move into position to pick up game pieces

c. Ability to move into position to deploy minibot

Efficient Game Piece placement
Mechanically easy delivery: We currently have a number of prototype manipulators being designed to deliver game pieces in the most reliable fashion possible. We are planning to choose one this week, but finals are kind of in the way <.<

Ability to deliver reliably in the top row: We currently plan to use a three-segment continuous elevator system with the manipulator mounted in front. By coincidence, our elevator interface is simple and modular, and will be able to support whatever manipulator we choose.

Effective use of sensor suite for approach and placement: Currently we are exploring the idea of “Autonomous Placement” (AP). The drivers of a robot with AP capability would be able to input some basic data parameters such as which line it is approaching and on which peg to place the game piece. Barring interference from other robots, the robot would then be able to reliably place the game piece without further driver interaction.

Interference from other robots is, of course, likely, so our placement will rarely be fully autonomous, but developing such an advanced sensor suite will be an invaluable resource for our drivers.

Efficient game piece pickup and control
From Feeder Station: Most of our manipulators involve some tricky placement by the feeder. A general theme in several of our manipulators is to place the logo piece on the prong horizontally and then let it fall into the vertical position. This can be a major problem, as some of our manipulators have very small tolerances as to the distance the robot can be from the feeder station to successfully receive a piece; a little too close or too far, and the logo simply falls off. It may be very difficult to reach that “sweet spot,” as the drivers are all the way across the field, and we aren’t certain how effectively our sensor suite can mitigate this risk. The design team should take this problem into careful consideration when choosing our manipulator.

From ground: Some of our manipulators can be adapted to pick up logo pieces from the ground, others cannot. Ground pickup is not a necessary robot feature; there are 36 logo pieces on the field, feeder stations most likely will not run out until late in the match.

Torque/traction
The reasoning behind having a lot of pushing power is that if a robot tries to block us from entering our alliance zone, we can just push them into our alliance zone instead, forcing them to GTFO our receive a penalty. In the process, they must get out of our way. This will also benefit us in the event that our manipulator flops; we will still be able to play solid defense by pushing/pinning other robots.

We opted for a six-wheel drive chassis rather than a mecanum drive because of the simplicity and power of the six-wheel drive model. Additionally, we chose to use plaction wheels on our central and rear wheels, sacrificing some turning effectiveness in favor of extra traction. While having four traction wheels on the ground at the same time will reduce our turning, we feel that the small drive base will mitigate most of the negative effects. If it is too much of a problem, we plan to put regular AndyMark traction wheels on the back instead. We will retain the traditional omni-wheels in the front. With regards to power, it is imperative that our center of gravity be as low as possible, even when lifting, so that we can push defending robots out of the way. A light manipulator is therefore most desirable with respect to this priority.

Fast/efficient minibot deployment
This is going to be majorly dependent on our sensor suite. Priorities are reliability and speed, we should be able to drive up to the tower, deliver the bot, and have it moving with 9.7 seconds remaining. We have been focusing on our logo placement for now, but this week we will give serious thought to our minibot deployment strategy, before we finalize our elevator/manipulator design.

Speed
Most of our design choices are conducive to a high-speed robot, we haven't really needed to devote a lot of attention to this category.

Maneuverability
Ability to move into position to score pieces: Our chassis has made some sacrifices in this area in order to further our progress in other areas. We gave up on a mecanum drive design because of its significant lack of torque and because it would make Zuhair’s job of implementing AP horribly complicated. We also replaced our rear omni wheels with plactions, which will slightly inhibit our turning. Maneuverability is a low priority for our robot because of our “through rather than around” drivetrain policy, but once in the alliance zone we will need to be able to maneuver to place game pieces on the pegs – ideally, AP should make this easy from a driver standpoint, but our robot still has to be maneuverable enough to actually make the play. This shouldn’t be too much of an issue – a 6WD robot can turn on a dime, and since we won’t be interfered with in our alliance zone, we can essentially go where we want once there. Overall, our mediocre maneuverability won’t hinder us from effective scoring.

Ability to move into position to pick up pieces: This shouldn’t be an issue for picking up pieces from the feeder station, but our six-wheel drive will make it difficult to get into position to lift from the ground, particularly if our manipulator’s “window” for successful pickup is small and our drivers cannot effectively use the camera.

Ability to move into position to deploy minibot: This is largely dependent on which deployment scheme we choose, but our 6WD should be adequate.

Conclusion
Because of our interpretation of this year's rules, our robot is an OFFENSIVE-minded robot that plans to go THROUGH, NOT AROUND defenders in order to score as EARLY AND OFTEN as possible, on the TOP ROW. I'm hoping to get some feedback from some people with more experience than I, particularly before we begin manufacturing our chassis. I also hope that some of you take away some lessons of your own. Thanks a lot for reading this WoT, it was five pages in OpenOffice.org, single spaced.

Here's to a great season,
Dolan “SirTasty” Murvihill
Team 2374

[Chris is me]

Wonderful job. This is a very thorough game analysis that all teams should be doing, and I couldn't find fault with anything in it just skimming the text.

That said, I think you're vastly understating the value of floor pickup.

[Justin Montois]

This is a very interesting strategic analysis. I agree with most parts but some are up for discussion I think. Looking at some of your underlaying principles...

1. Focusing on defense is an INFERIOR strategy

This is true to an extent, but I really REALLY don't agree with the logic behind "If we can't play offense well we can always play defense" I know that if I designed a robot to play defense it will do it better then your offensive robot that "can" play defense. Look at this thread.. http://www.chiefdelphi.com/forums/sh...threadid=89352

That robot has a great chance to shut down any great offense robot. I don't think you can say that focusing on defense is inferior, if you design a robot to play defense and you prevent your offensive juggernaut from even getting to the scoring rack who has the better strategy?

2. One minibot will not perform significantly better or worse than another

I think you will be surprised at competition. I think there will be several average minibots that don't perform significantly better then the others, but there will be one or two that blows everyone else away. This remains to be seen.

3. Teams have the greatest opportunity to affect the outcome of the match if they focus on hanging game pieces.

This is another tough call IMO. I think a great defensive robot, or better yet a defensive ALLIANCE that completely barricades the scoring zone for the opposing alliance but has 2 amazing minibots can eliminate your offensive alliance without scoring a single tube.

I think there will be teams playing a lot of defense at regionals that will give the pure scorers a run for their money. I just watched about 10 match videos from 2007, unless teams across the board improve their manipulators, a couple good hits from an opposing alliance robot would cause the robot to drop their tube. Make a team drop their tube enough and they will realize it's easier to stop another team from scoring a tube then it is for them to score one themselves. By Saturday more teams are playing defense and it's getting harder to score.

This is what makes strategy in FIRST so fun IMO. Your strategic analysis might be spot on, or it may not be. After regionals start and you see how the game is being played, you watch it evolve over the weeks and at Championship everyone knows their strategy and the games begin.

Best of luck this season.

[Josh Goodman]

Very nice Sir, I agree with Justin, this is the best part of FIRST. Because not only does the game effect what is most important, but the other teams do as well which is why from year to year it's so different.

Our team did a very similar strategy breakdown the first day as well. Yours is very neat and organized. I wish you the best of luck in this years game and hope your robot is built to be everything you wanted!

[Brandon Holley]

Great game analysis, but I need to agree with Chris here on your views of ground pickup.

It would really suck to try and place a piece on the scoring pegs, only to drop it, and then have to run all the way across the field to get a new one and try again. For this reason alone, ground pickup should be a bigger consideration.

Great job overall though!

-Brando

[pfreivald]

Originally Posted by Brandon Holley

Great game analysis, but I need to agree with Chris here on your views of ground pickup.

x1000.

Great analysis overall, but I think ground pickup is an essential aspect of any competitive robot in this game. Whether it's your own piece that fell, or an alliance partner's, or an opponent's -- if it's on the ground and it's closer to your scoring rack than your feeders, you want to be able to get it and score it.

[EOC]

Great job! I agree with almost everything. I think that some minibots will be faster than others.

[SirTasty]

Originally Posted by 340x4xLife

1. Focusing on defense is an INFERIOR strategy

This is true to an extent, but I really REALLY don't agree with the logic behind "If we can't play offense well we can always play defense" I know that if I designed a robot to play defense it will do it better then your offensive robot that "can" play defense. Look at this thread.. http://www.chiefdelphi.com/forums/sh...threadid=89352

That robot has a great chance to shut down any great offense robot. I don't think you can say that focusing on defense is inferior, if you design a robot to play defense and you prevent your offensive juggernaut from even getting to the scoring rack who has the better strategy?

In my experience, defense is much easier than offense. Last year our robot was built for offense, and it turned out that our manipulator sucked but it was the best defensive robot at our regional. Look at it this way: a robot that is built for scoring but bad at it can still play defense. A robot that is built for defense but is mediocre at it will never play good offense. The chassis and drivetrain we have chosen are very conducive to strong defense.

Yes, I saw 3553's strategy, and liked it, but I think our drive train will have the advantage there. With four CIMs and four plaction wheels pushing at the edge of their robot, we should be able to push them into our alliance zone; this won't cause them to incur a penalty under <G59>, but they will need to take all necessary steps to leave it immediately, which will most likely include getting out of our way.

I just don't see that any robot will be able to play a 100% lockout defense in this year's game.

Originally Posted by 340x4xLife

2. One minibot will not perform significantly better or worse than another

I think you will be surprised at competition. I think there will be several average minibots that don't perform significantly better then the others, but there will be one or two that blows everyone else away. This remains to be seen.

FIRST has deliberately designed the minibot game to be as boring as possible. I just don't see that it would be that good.

Originally Posted by 340x4xLife

3. Teams have the greatest opportunity to affect the outcome of the match if they focus on hanging game pieces.

This is another tough call IMO. I think a great defensive robot, or better yet a defensive ALLIANCE that completely barricades the scoring zone for the opposing alliance but has 2 amazing minibots can eliminate your offensive alliance without scoring a single tube.

I think there will be teams playing a lot of defense at regionals that will give the pure scorers a run for their money. I just watched about 10 match videos from 2007, unless teams across the board improve their manipulators, a couple good hits from an opposing alliance robot would cause the robot to drop their tube. Make a team drop their tube enough and they will realize it's easier to stop another team from scoring a tube then it is for them to score one themselves. By Saturday more teams are playing defense and it's getting harder to score.

I'll answer this point later, I'm being kicked out of the library for finals.

Originally Posted by 340x4xLife

This is what makes strategy in FIRST so fun IMO. Your strategic analysis might be spot on, or it may not be. After regionals start and you see how the game is being played, you watch it evolve over the weeks and at Championship everyone knows their strategy and the games begin.

Best of luck this season.

[Bill_B]

Thanks a lot for reading this WoT, it was five pages in OpenOffice.org, single spaced.

Why not put this into papers section?

[AllenGregoryIV]

This was a very good analysis but I will say that my team came to several different conclusions.

Through, not Around?
One thing I see is that you didn't mention your actual gearbox. You are planning on having power and speed but don't mention if you are planning on having a 6 motor 2-speed drive train or just 4 CIMs geared to a reasonable 10ft per sec (or something along those lines). The latter will not allow you push the most powerful robots and it also won't let you get back and forth across the field faster than the fastest robots.

I find it interesting that a lot of teams seem to be going for the go through not around strategy. It is still not clear how long a defender has to get them self out of your zone after they have been pushed in. What if they flip over while being pushed and now you can't score on that side the rest of the match? I also think you have to be cautious of getting pinning calls against you if you are trying to push through someone that happens to be touching a lane divider or a tower.

Every second is going to count when hanging tubes; going from 20sec a tube to 15sec a tube is 2 more tubes a match, not counting minibot time. So having to drive through someone that is slowing you down is going to take precious time away from actually scoring tubes.

For these reasons, I think teams that are building for the go around not through strategy will have success this year. (Teams that can do both, i.e. swerve drive, will fair even better). Even if you don't have a powerful drive train the field elements become your friend, running to them makes it so no other robot can touch you without starting a pinning count.

Floor Loading
I agree with the others about floor pickup. What if your feeder drops a tube right in front of the feeder station? Will your robot be able to get any more tubes? Not being able to pick up from the floor will make for a defenders dream because they can just push tubes (think about un-hung ubertubes) into your lane and it will be hard for you to remove them.

Defense
I agree with you about not designing for defense. Defense is going to be harder than people think this year. The longest any defender can be is 7 feet. The zone is 18 feet wide. If two offensive robots are trying to score one of them just has to set a pick for the other and you can easily both get into your zone and be safe. (Yes if three robots are all defending a zone it will be impossible to score, they better keep that up for the entire match including the MINIBOT race, because otherwise you have two robots who have had more time to setup for the race and one that is capping an ubertube for 6 more points)



I wish everyone the best of luck with their strategist and will see how they all play out; I think this going to be a really exciting game and one of the most intense for the strategist on every team.

[IndySam]

Nice job, my thoughts.


Floor loading will be critical.

Don't focus on bottom pegs, you would be better off delivering tubes to an alliance partner than hanging tubes on the bottom row.

That being said a defensive robot that can also deliver tubes could be a nice addition to any alliance, at least at regionals.

[Lil' Lavery]

You went from point A to point C without venturing through point B.

You analyzed the game and came up with some very basic ideas of what features are important and how they might impact your chances of winning. But you never focused on how to win. That is an essential step in building a winning strategy, both for an alliance and for your robot. Specifically, how your robot fits into an alliance.

This doesn't automatically mean the conclusions you reached are wrong, though.

[mathking]

I like the thoroughness of the overall assessment. From all of the discussion here on CD, one thing has really jumped out at me this year. Everyone seems to be decrying scoring low as a decent option, but I think that those who are doing this are not thinking about reliability and speed.

Consider:
A logo across the bottom row is worth 6 points, the same as two pieces across the top. If you assume a 1 in 3 rate of dropped tubes (and I will bet it will be higher than that for many teams) and a reasonable time difference for the drop - pick up off the floor - rehang operation between robots hanging low and robots hanging high, all of sudden the difference in expected values narrows considerably.

I still agree that hanging on the top row is vital to be a top team, but I also think that teams should not ignore the bottom.

[flaminwhale]

Has anyone thought of trying to catapult a tube over a robot that is trying to prevent you from getting into your alliance zone?

just a thought

[SirTasty]

Lots of replies, I'm glad to see all the feedback! Just a few comments for folks here.

Originally Posted by 340x4xLife

This is a very interesting strategic analysis. I agree with most parts but some are up for discussion I think. Looking at some of your underlaying principles...

1. Focusing on defense is an INFERIOR strategy

This is true to an extent, but I really REALLY don't agree with the logic behind "If we can't play offense well we can always play defense" I know that if I designed a robot to play defense it will do it better then your offensive robot that "can" play defense. Look at this thread.. http://www.chiefdelphi.com/forums/sh...threadid=89352

That robot has a great chance to shut down any great offense robot. I don't think you can say that focusing on defense is inferior, if you design a robot to play defense and you prevent your offensive juggernaut from even getting to the scoring rack who has the better strategy?

I saw that thread and I was impressed by the innovation in 3553's strategy, but as Gregory pointed out, the longest they can be is eight feet long. The alliance zone is eighteen feet wide. Even two of these robots won't be able to block the entire zone, and in any case I think we'll have the power to push past them, or push them into our alliance zone. I just don't think that robot will be as viable as it looks at first glance, and I don't think any defensive robot can completely stop a strong offensive team.

Originally Posted by 340x4xLife

2. One minibot will not perform significantly better or worse than another

I think you will be surprised at competition. I think there will be several average minibots that don't perform significantly better then the others, but there will be one or two that blows everyone else away. This remains to be seen.

I just don't see how this could be done under the current minibot rules. Perhaps there's something I missed; I guess we'll find out at regionals.

Originally Posted by 340x4xLife

3. Teams have the greatest opportunity to affect the outcome of the match if they focus on hanging game pieces.

This is another tough call IMO. I think a great defensive robot, or better yet a defensive ALLIANCE that completely barricades the scoring zone for the opposing alliance but has 2 amazing minibots can eliminate your offensive alliance without scoring a single tube.

Even two defensive robots cannot completely barricade the scoring zone, and three defensive robots would never score themselves. Even as the game reaches the finale, the offensive team is between the defensive team and their towers (and thus, their winning minibot strategy)..

Originally Posted by 340x4xLife

I think there will be teams playing a lot of defense at regionals that will give the pure scorers a run for their money. I just watched about 10 match videos from 2007, unless teams across the board improve their manipulators, a couple good hits from an opposing alliance robot would cause the robot to drop their tube. Make a team drop their tube enough and they will realize it's easier to stop another team from scoring a tube then it is for them to score one themselves. By Saturday more teams are playing defense and it's getting harder to score.

Keep in mind that during the most fragile part of the process, the actual act of hanging the tube, the robot is protected from all robot impact.

Originally Posted by Bill_B

Why not put this into papers section?

Because I am a noob.

Originally Posted by AllenGregoryIV

This was a very good analysis but I will say that my team came to several different conclusions.

Through, not Around?
One thing I see is that you didn't mention your actual gearbox. You are planning on having power and speed but don't mention if you are planning on having a 6 motor 2-speed drive train or just 4 CIMs geared to a reasonable 10ft per sec (or something along those lines). The latter will not allow you push the most powerful robots and it also won't let you get back and forth across the field faster than the fastest robots.

I find it interesting that a lot of teams seem to be going for the go through not around strategy. It is still not clear how long a defender has to get them self out of your zone after they have been pushed in. What if they flip over while being pushed and now you can't score on that side the rest of the match? I also think you have to be cautious of getting pinning calls against you if you are trying to push through someone that happens to be touching a lane divider or a tower.

We have four CIMs in two SuperShifters driving our wheels, we should be able to shift to high gear when we need to move, low when we need to push. We're also not too worried about flipping robots in our zone. Defensive robots tend to be built low.

Originally Posted by AllenGregoryIV

Every second is going to count when hanging tubes; going from 20sec a tube to 15sec a tube is 2 more tubes a match, not counting minibot time. So having to drive through someone that is slowing you down is going to take precious time away from actually scoring tubes.

For these reasons, I think teams that are building for the go around not through strategy will have success this year. (Teams that can do both, i.e. swerve drive, will fair even better). Even if you don't have a powerful drive train the field elements become your friend, running to them makes it so no other robot can touch you without starting a pinning count.

Going around takes much longer than going through.

<snip>

[quote=AllenGregoryIV;1003470]
Defense
I agree with you about not designing for defense. Defense is going to be harder than people think this year. The longest any defender can be is 7 feet. The zone is 18 feet wide. If two offensive robots are trying to score one of them just has to set a pick for the other and you can easily both get into your zone and be safe. (Yes if three robots are all defending a zone it will be impossible to score, they better keep that up for the entire match including the MINIBOT race, because otherwise you have two robots who have had more time to setup for the race and one that is capping an ubertube for 6 more points)
[quote]

That's a very succinct way of putting it and I agree 100%.

Originally Posted by Lil' Lavery

You went from point A to point C without venturing through point B.

You analyzed the game and came up with some very basic ideas of what features are important and how they might impact your chances of winning. But you never focused on how to win. That is an essential step in building a winning strategy, both for an alliance and for your robot. Specifically, how your robot fits into an alliance.

This doesn't automatically mean the conclusions you reached are wrong, though.

I don't see the hole here. How to win: score more points than the other guys. How to do that: stop the other guys from scoring, or score more. Since our first principle was "defense is inferior," it follows directly that our strategy is "score more." There are two ways of doing that: hanging or minibot. Since we don't expect minibots to be an aspect of the game that can be easily affected, it only follows that we must hang. Since we don't think the bottom row is important, we must hang on the top rows. That seems like a pretty good strategy, but this is the first year I've been deeply involved in strategy. Could you elaborate on where you see my error?

Originally Posted by mathking

I like the thoroughness of the overall assessment. From all of the discussion here on CD, one thing has really jumped out at me this year. Everyone seems to be decrying scoring low as a decent option, but I think that those who are doing this are not thinking about reliability and speed.

Consider:
A logo across the bottom row is worth 6 points, the same as two pieces across the top. If you assume a 1 in 3 rate of dropped tubes (and I will bet it will be higher than that for many teams) and a reasonable time difference for the drop - pick up off the floor - rehang operation between robots hanging low and robots hanging high, all of sudden the difference in expected values narrows considerably.

I still agree that hanging on the top row is vital to be a top team, but I also think that teams should not ignore the bottom.

Not so. TWO logos across the bottom row are worth twelve points. ONE logo across the top row is worth eighteen! Add an ubertube to that for even more fun. So, really you're looking at bottom hangers needing to hang twice as much in the same time in order to equalize. Top hangers have plenty of wiggle room inside their alliance zone.

regarding floor pickup: I am surprised at how many people are emphasizing floor pickup here. I am inclined to bow to your experience here.