Chief Delphi

Chief Delphi (http://www.chiefdelphi.com/forums/index.php)
-   Rules/Strategy (http://www.chiefdelphi.com/forums/forumdisplay.php?f=6)
-   -   Efficient Points (http://www.chiefdelphi.com/forums/showthread.php?t=154174)

Chris is me 25-01-2017 10:04

Re: Efficient Points
 
Quote:

Originally Posted by MrForbes (Post 1636060)
I know that shooting them is easy. The hard parts are getting them to the shooter quickly and consistently, and aiming at the goal so most of the balls go in.

To some extent, the faster you shoot, the less accurate you have to be for that shot to have been worth it. An extreme example: If you shoot 100 balls a second and 20% of them go in, that's much better than shooting 5 balls a second and 80% of them going in, particularly when game pieces are plentiful, you can sometimes catch your own misses, and you take shots quickly in between gear scoring runs.

No one should read this as "don't focus on accuracy", but the accuracy requirements for this game, versus 2016's single ball shot, are vastly different.

MrForbes 25-01-2017 10:10

Re: Efficient Points
 
Shooting 5 balls a second, and 80% of them going in, would be just fine. Will your team be doing this? Ours isn't going to even try.

ToddF 25-01-2017 10:25

Re: Efficient Points
 
Quote:

Originally Posted by WhiteBoySwag (Post 1635584)
So after looking into how to score points and where we get our game pieces from, I haven't found a viable use for fuel when relaying gears the whole competition would most likely be a much better use of time.

Our goal is to score 4 RP in our qualification matches. Breaking it down:
- 1RP from 40kPA fuel
- 1RP from 12 gears
- 2RP from winning a match

The first ranking point can be earned by a single robot acting alone. (Reward for individual performance) Barring defense, which is often not a factor in qualification matches, the second ranking point can be earned by an alliance. (Reward for team performance) The final 2 RP not only depend on your own alliance, but on how the opposing alliance performs. Barring defense, this is not within our control.

It makes strategic sense to prioritize scoring actions in that order. First get the RP you can earn by yourself. Then get the RP you can earn with good partners. Performing those tasks will, more often than not, earn you the final 2 RP. In the final matches of an event, where alliances are more evenly matched, you may need to modify your strategy to take actions which effect the opponents ability to score. But, at that point it's smart gameplay strategy and robot reliability, not robot conceptual design, which makes the difference.

So, making my own predictions for the season: 90% (or higher) of matches where 4RP are earned will have the 40kPa RP scored in the autonomous period. Also, 75% (or higher) of matches on Einstein will have a 40kPa auto period. At least 6 of the 8 Einstein alliances will be led by teams which are capable of a 40kPa auto mode.

The ultimate Einstein alliance has a robot with a 40kPa auto mode, floor fuel pickup, and average gear handling, paired with two extremely fast gear cyclers. While the first robot may not be within the reach of many teams, the latter definitely is. If you are going for a gear handling bot, be sure to be far above average at that task. If your ambition is to be a championship alliance leader, you better be working toward a 40kPa autonomous mode.

Chris is me 25-01-2017 10:26

Re: Efficient Points
 
Quote:

Originally Posted by MrForbes (Post 1636081)
Shooting 5 balls a second, and 80% of them going in, would be just fine. Will your team be doing this? Ours isn't going to even try.

High goal shooting is less of a priority for my team versus many others - our software is limited, we don't have the manpower to scout super duper well, and drive practice is our strength, so it's not the part of the game that best aligns for us. So it depends on how prototyping continues to go for us. The actual high goal shooter is the least critical of the main things the robot might do, so we have not focused super hard on increasing throughput until recently.

But I really see this game playing like 2009, where volume of scoring is the important thing more than accuracy. There are six hundred balls to work with, and you have another time consuming task that you will have to help with on any alliance (gears), so you want your shooting to be quick opportunistic points that occur as a thing you do in between gear cycles. In 2009 the very best (one ball wide) shooters could get six balls per second, with a game piece much harder to transfer energy to, and much less motor power available. I'm hoping as a stretch goal, if everything goes right, to double that.

MrForbes 25-01-2017 10:49

Re: Efficient Points
 
We did pretty well in 2009, once we aimed our shooter down instead of up. That was an easier game as far as designing the robot---the robot only had to do one function, which was to move balls as fast as possible. This year there are three things the robot has to do. With a team of mostly inexperienced students, we decided to try for the two that have the highest chance of success.

I think we scored one high goal last year. Reality is tough.

Chris is me 25-01-2017 10:55

Re: Efficient Points
 
Quote:

Originally Posted by MrForbes (Post 1636105)
We did pretty well in 2009, once we aimed our shooter down instead of up. That was an easier game as far as designing the robot---the robot only had to do one function, which was to move balls as fast as possible. This year there are three things the robot has to do. With a team of mostly inexperienced students, we decided to try for the two that have the highest chance of success.

I think we scored one high goal last year. Reality is tough.

Not focusing on shooting at all is a super reasonable thing to do this year; I was just explaining how we were going about shooting if we get enough time to actually implement it. I figured that's what you were asking about, the details of that and not the details of shooting at all.

Ginger Power 25-01-2017 11:01

Re: Efficient Points
 
Quote:

Originally Posted by Chris is me (Post 1636092)
There are six hundred balls to work with, and you have another time consuming task that you will have to help with on any alliance (gears), so you want your shooting to be quick opportunistic points that occur as a thing you do in between gear cycles.

This is exactly why 4607 has what is essentially a 2 stage hopper. The first stage is large (60 Fuel) but the transfer to the shooter is relatively slow. The second stage is small (24 Fuel) but the transfer to the shooter is extremely rapid. This setup will allow us to empty the second stage hopper during every Gear cycle. Then while we're in the process of cycling our next gear, we can be transferring Fuel from our slower stage 1 hopper to our fast stage 2 hopper. This setup should minimize the amount of wasted time hopefully!

JesseK 25-01-2017 11:19

Re: Efficient Points
 
Quals:
  • Suppose a gear bot and a fuel bot have a 50% (equal) chance to win a match.
  • In their winning matches, the average gear bots will win with more points vs the average ball bot.
  • If they're likely to get more points, they're likely to win more than 50% of matches vs bots who do balls
  • Therefore ball & gear bots will likely net the same RP in a 1-to-1 comparison.
  • Therefore gear bots likely rank higher than ball bots, all else equal (reference 2nd-order sort this year)
  • Since Qual schedules are random and not equal, seeding will likely be greatly determined by schedule difficulty, which is impossible to predict at this time. Yet securing as many points as possible solo will net the highest possible chances of seeding high.
  • Therefore Gear bots MUST hang in order to increase probability of winning in a random schedule and keep a 1-to-1 RP ratio with fuel bots that do not always win. It's not like gear bots have anything valuable to do with gears in the last 5 seconds anyways.

Elims:
  • 4 rotors are not a given, even at champs. Concerted effort must be put into it, and if all effort is given to it then there is likely little time for much else.
  • There are easy strategic scenarios against any alliance that employs a ball specialist since 4 rotors are even less likely than with a 3-gear alliance (outside of 0.2% of teams).
  • There are few strategic scenarios that that will win against an all-gear alliance where the best two gear bots at an event have paired up. Even at champs.
  • This ignores the 0.2% of teams (approximately 60 worldwide) who will find a way to net the required throughput and accuracy in order to outscore gears on a 'per-trip' basis.
  • Gear specialists will likely find a home with the top 0.2% of teams since it means that 0.2% of teams can focus on their role.

Thus for 99.8% of teams, gears give more efficient points.

D.Allred 25-01-2017 11:54

Re: Efficient Points
 
Quote:

Originally Posted by JesseK (Post 1636122)

...Thus for 99.8% of teams, gears give more efficient points.

Climbing is the most efficient - although a single high value task. You'll need gears and climbing to play with the 0.2%.

David

Ginger Power 25-01-2017 12:34

Re: Efficient Points
 
Quote:

Originally Posted by JesseK (Post 1636122)
Thus for 99.8% of teams, gears give more efficient points.

I agree with this line of thinking if teams are forced to decide between Fuel OR Gears. But since teams can choose to do both, I think the numbers will work out differently for many teams.

I'm sure many teams are going to do what I call "Super Cycles" where they score a Gear and Fuel in the same repeatable sequence. For these teams I suspect they will choose to intake Fuel while driving to and from the Retrieval Zone and after they are finished depositing their Gear, they'll make a quick pit stop in the Key to score their collected Fuel. The time it takes to do this pit stop will determine the effectiveness of the Super Cycle vs. cycling Gears only. If things work out the way I'm thinking they will, teams that do Super Cycles will be more efficient (and a more valuable alliance partner) than teams that only have Gear cycling capability.

Donut 25-01-2017 12:37

Re: Efficient Points
 
Quote:

Originally Posted by JesseK (Post 1636122)
Quals:
  • Suppose a gear bot and a fuel bot have a 50% (equal) chance to win a match.
  • In their winning matches, the average gear bots will win with more points vs the average ball bot.
  • If they're likely to get more points, they're likely to win more than 50% of matches vs bots who do balls
  • Therefore ball & gear bots will likely net the same RP in a 1-to-1 comparison.
  • Therefore gear bots likely rank higher than ball bots, all else equal (reference 2nd-order sort this year)

Points 1 and 3 contradict each other. If gear bots as a whole are likely to win more than 50% of matches vs fuel bots, then either gear bots had a >50% chance of winning matches or fuel bots had a <50% chance to win matches (or both).

Additionally, I don't see how point 2 leads to the conclusion of point 3. How do we conclude margin of victory in winning matches predicts winning more total matches, unless you are predicting a higher average points scored per match for gear bots (wins and losses), which means a greater probability of winning matches in the first place (counter point 1)?

JesseK 25-01-2017 13:20

Re: Efficient Points
 
Quote:

Originally Posted by Ginger Power (Post 1636140)
I agree with this line of thinking if teams are forced to decide between Fuel OR Gears. But since teams can choose to do both, I think the numbers will work out differently for many teams.

I'm sure many teams are going to do what I call "Super Cycles" where they score a Gear and Fuel in the same repeatable sequence. For these teams I suspect they will choose to intake Fuel while driving to and from the Retrieval Zone and after they are finished depositing their Gear, they'll make a quick pit stop in the Key to score their collected Fuel. The time it takes to do this pit stop will determine the effectiveness of the Super Cycle vs. cycling Gears only. If things work out the way I'm thinking they will, teams that do Super Cycles will be more efficient (and a more valuable alliance partner) than teams that only have Gear cycling capability.

I suspect that an above-average gear bot is capable of a gear cycle in the time it takes an above-average ball bot to offload a full hopper. Whether the gears are worth it depends on whether the 'super cycle' bot will get the 3rd rotor or not, and what happened to its autonomous gear. Given equal resources and time to develop, I suspect the gear bot will have a more reliable 65-pt gear auton and can also reliably solo up through the 3rd rotor*. Disregarding hanging, a super-cycle bot must be able to hit 145 points with auton and 'super cycles'. This is doable, but IMO isn't what an above average team will produce. At that point it would come down to how the alliance would help.

*so long as partners at least drop their auton gears, this is likely so trivial a good gear bot can use the extra time to play defense on the super cycle bot

Quote:

Originally Posted by Donut (Post 1636143)
Points 1 and 3 contradict each other. If gear bots as a whole are likely to win more than 50% of matches vs fuel bots, then either gear bots had a >50% chance of winning matches or fuel bots had a <50% chance to win matches (or both).

Additionally, I don't see how point 2 leads to the conclusion of point 3. How do we conclude margin of victory in winning matches predicts winning more total matches, unless you are predicting a higher average points scored per match for gear bots (wins and losses), which means a greater probability of winning matches in the first place (counter point 1)?

Point 2 makes an assertion that in matches where a gear threshold was met (3rd or 4th rotor) that the win margin will be much greater than for the losing matches when the gear threshold was not met. It means that this game is about hitting the marks for the rotors, and win/loss will likely be determined by who can do that more reliably.

Point 3 does contradict Point one, but Point 1 was a supposition rather than an assertion. Your second paragraph is correct - I presume that more points = more wins, and since starting the next rotor means many more points than an average cycle of balls into the boiler, it is likely a gear bot will win more. Sorry this wasn't clear.

Ginger Power 25-01-2017 14:17

Re: Efficient Points
 
Quote:

Originally Posted by JesseK (Post 1636159)
I suspect that an above-average gear bot is capable of a gear cycle in the time it takes an above-average ball bot to offload a full hopper. Whether the gears are worth it depends on whether the 'super cycle' bot will get the 3rd rotor or not, and what happened to its autonomous gear. Given equal resources and time to develop, I suspect the gear bot will have a more reliable 65-pt gear auton and can also reliably solo up through the 3rd rotor*. Disregarding hanging, a super-cycle bot must be able to hit 145 points with auton and 'super cycles'. This is doable, but IMO isn't what an above average team will produce. At that point it would come down to how the alliance would help.

*so long as partners at least drop their auton gears, this is likely so trivial a good gear bot can use the extra time to play defense on the super cycle bot

To get three Rotors and alliance needs to deliver 6 total gears to the Airship. Lets assume an above average Gear cycler can do 6 Gears on its own (maybe a little high but I think reasonable), and that average Gear cyclers can do 2-3 Gears. I also think you're probably right that an above average Super Cycle robot will take about the same amount of time to empty their hopper as a Gear specialist will take to complete one full Gear cycle. So we can then assume a Super Cycle robot will do 3 Gears and 3 Hoppers.

In the case of the Gear specialist, assuming you're paired with 2 average robots, you'll be able to deliver 6+(2 or 3)+(2 or 3) = 10 - 12 Gears. This means that you'll be heavily dependent upon your alliance partners to turn 4 Rotors, and you will be heavily susceptible to defense that slows you down just enough so that you just miss 4 Rotors.

I think in situations such as this where you know delivering 12 Gears will be close, you're better off just planning on delivering 6 Gears, and spending the rest of the time doing other valuable tasks like defense, or scoring Fuel (which isn't possible in the case of the Gear only robots). If I'm playing against an alliance that might just barely score 12 Gears, I'm going to play extreme defense towards the end of the match to ensure you just miss your 12th Gear.

In the case of the Super Cycle robot, continuing the assumption that you're with 2 average robots, you'll be able to deliver 3+(2 or 3)+(2 or 3) = 7 - 9 Gears. I personally like this situation much better. You have a decent buffer against defense/mistakes and you are also shooting 3 hoppers worth of Fuel into the Boiler. The value of this Fuel will be completely dependent on the effectiveness of the shooter, but some Fuel points are basically guaranteed. These Fuel points will more often than not be the difference in matches where both alliances turn 3 Rotors and have the same number of climbs.

Disclaimer: Obviously this analysis is simplified and doesn't include factors like autonomous. The analysis also changes if you assume the average robot can cycle more than 2-3 Gears. In this case, Gear specialists will likely be able to turn the 4th rotor more consistently and in doing so will be a much more sensible design choice.

TL;DR: Designing your robot to handle solely Gears is risky because you're dependent upon alliance partners to achieve a Rotor advantage vs. the opposing alliance.

JesseK 25-01-2017 14:49

Re: Efficient Points
 
Quote:

Originally Posted by Ginger Power (Post 1636184)
TL;DR: Designing your robot to handle solely Gears is risky because you're dependent upon alliance partners to achieve a Rotor advantage vs. the opposing alliance.

The tricky thing with comparison, and you mentioned it, is autonomous. Gear specialists can (probably) count on 3 easy gears (plus the spare). Non-specialists can't, unless they find SWAP for a gear ground intake.

You make a good point though. It seems like the MCC for a gear specialist is a reliable auton and then at least 3 full-field gear cycles under even the heaviest of defense (2v1). Other than that, I'm not sold that it's riskier than over-extending a team's capabilities by doing more mechanisms and dividing attention during an event. This bias is based upon my team's history of over-extending.

With an all-gear Quals alliance as you described, I'd probably start the match with an understanding that we would attempt 4 rotors and adapt from there with preset milestones and knowing who would play what defense if a call was made to abort. The gameplay strategies (and a little luck in execution) are the secret sauce for making 4 rotors work in that situation, so I won't reveal too much ;).

Ginger Power 25-01-2017 15:16

Re: Efficient Points
 
Quote:

Originally Posted by JesseK (Post 1636200)
Other than that, I'm not sold that it's riskier than over-extending a team's capabilities by doing more mechanisms and dividing attention during an event. This bias is based upon my team's history of over-extending.
[snip]
The gameplay strategies (and a little luck in execution) are the secret sauce for making 4 rotors work in that situation, so I won't reveal too much ;).

I completely agree that over-extending is a killer. A lot of my arguments aren't taking into consideration the difficulty that a Fuel scoring mechanism adds to a robot. Scoring Fuel, Gears, and Climbing is significantly harder than just Gears and Climbing. Teams who focus on the latter are undoubtedly going to have more driver practice than teams who attempt the former.

I think this year the "cool factor" of a wiffle-ball-machine-gun-robot is going to hurt the competitive level of a large portion of FRC teams. The arguments I'm making apply to the relatively few teams that can effectively pull it off.

On another note, you have me thinking about what strategies you could employ to counteract a defensive blitz at the end of the game. My first thought was that you could just shuttle Gears to the base of your Airship during the early game, ejecting all the Gears on the ground near your springs. Then during the late game defensive blitz, you could just sit by your Airship and use your Gear ground pickup to quickly score all the Gears you cycled earlier in the match. High risk, high reward strategy... sitting on a bunch of Gears for most of the match would be pretty terrifying.


All times are GMT -5. The time now is 08:45.

Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
Copyright © Chief Delphi