A statistical analysis of the "autonomous advantadge"

[Greg Marra]

Update:

I have created sheets so that anyone attending a Regional can help collect data. Simply watch the match, and write down the information the sheet asks for. Don't feel obligated to do every single match. If you are only able to do a handful, it is better than nothing. I just ask that you don't record practice matches, since they are often played very differently than real matches.

Here is a nice printable version of the sheet. There are twenty rows per sheet, so you will probably need about 4 or 5 to completely cover a regional. If you really want a .xls version, PM me and I'll send it to you.

Once you have filled out a sheet either scan it, take a digital photo of it, or type it back into a spreadsheet. Then e-mail it to: aim.high.autonomous [at] gmail [dot] com. I'll take everyone's data, compile it, make it available, and then run some statistics on it.

[/update]

Today, while observing the competition at BAE, there was something odd occurring. In the elimination matches, whoever won autonomous won the round. Every time. Fourteen of fourteen times (excluding the ties). This seems to be telling.

However, how can we tell if it is winning autonomous that is actually the root of causation here? There are many other variables. If a team has a working autonomous mode, maybe they just have a better team and a better robot. If a team wins autonomous, is it the ten point bonus plus the points they score that gives them the game? Is it possible that going on defence first is such a large advantage that the other alliance cannot come back to reverse the tide?

This is an interesting question that deserves study. However, you can not study a question without data. Here is my call:

Either by analyzing Week 1 video or by collecting information at Week 2, we can carefully determine what is the deciding factor here.

The information of interest are: What are the scores following the end of autonomous (including the 10 point bonus)? Which alliance went on offence first? What are the scores at the end of the match (and how many came from ramping and penalties)? Was this match a qualifier, or an elimination match?

This will let us carefully separate out to see exactly what the underlying cause is. If it is a point differential, we can compare the lead at the end of auto to the lead at the end of the match. If it is going on defense first, then analyzing matches where autonomous was tied and defense is randomly selected, we will be able to deduce this.

Is anyone else interested in helping me collect data to analyze this? I think the answer might be interesting, and could help everyone gain a better understanding of the game.

[Lil' Lavery]

While it wasn't AS LARGE of a choakhold at VCU, the winning autonomous alliance typically did win matches, especially in the eliminations (and watching 1731 finally lock on with their turrent and drain 6 shots was quite impressive!)

[ldeffenb]

Originally Posted by Lil' Lavery

While it wasn't AS LARGE of a choakhold at VCU, the winning autonomous alliance typically did win matches, especially in the eliminations (and watching 1731 finally lock on with their turrent and drain 6 shots was quite impressive!)

Videos of VCU's finals are being posted (see http://www.chiefdelphi.com/forums/sh...d.php?t=45009). However, they seemed to have quit posting the score visually and I didn't take the time in some of the matches to find the audio announcement, so you might have to look elsewhere to find the final score. Sorry 'bout that!

Lynn (D) - Team Voltage 386 Software

PS. I'm anxious to see the outcome of the analysis. We are planning to WIN autonomous in our matches this coming weekend in the Central Florida Regionals. That is, IF the mechanical folks can get our ball storage and feeder working!

[Lil' Lavery]

Originally Posted by ldeffenb

PS. I'm anxious to see the outcome of the analysis. We are planning to WIN autonomous in our matches this coming weekend in the Central Florida Regionals. That is, IF the mechanical folks can get our ball storage and feeder working!

I don't know many teams that planned to lose it.....

[Nuttyman54]

Originally Posted by ldeffenb

they seemed to have quit posting the score visually and I didn't take the time in some of the matches to find the audio announcement, so you might have to look elsewhere to find the final score. Sorry 'bout that!

The final scores can be found on the FIRST website.

[Ryan F.]

Originally Posted by Nuttyman54

The final scores can be found on the FIRST website.

Amusing...apparently there must be some really really good shooters .

From First Website-

Q11 3/2/2006 11:40:00 AM
804 11106
384
540
1610 7

[George A.]

I'd say winning autonomous by either means (either winning it by points, or by getting the randomization in the event of a tie) is a HUGE advantage. Because most teams will unload their 10 preloaded balls in their autonomous mode, so whoever loses autonomous then has to spend time reloading balls. Whereas the winner of the autonomous can reload and be ready to go during their scoring phase.

I can't tell you how many times (as an arbitrary example) red would win and go reload, blue lost and would spend so much time reloading that they can't get a shot off. Meanwhile the red team is already set up and as soon as the whistle sounds red fires up their shooters before blue can do anything about it

Even if you don't get the 10 pt bonus, the way the match is set up, using that 40 seconds for 80 seconds of non stop scoring is unbelievable.

[Joel J]

Alright.. I have only observations to go on. You can all collect the data, together, and we can compare sometime later.

I'm saying this after watching VCU.

- When both alliances scored zero in autonomous, the randomly selected "defense" alliance still seemed to come out on top.

- If you win in autonomous, then you most likely have a good shooter, good software, good 1pt dumper, or a decent mixture. That is, if you are making it happen in autonomous, then you have what it takes to make it happen in the rest of the match.

- Teams on defense first get time to load up their robots with balls (either from the floor or human player) without cutting into their scoring time. They are then able to align themselves for shooting before their turn on offense begins allowing them to score points BEFORE their opponents can get into a defensive posture. Teams on offense after autonomus are usually close to their opponents (giving their opponents perfect positioning for playing defense), far away from the 3pt goal, and have very few balls.

- With the amount of points I saw being scored, the 10 point bonus gained from winning autonomous is significant. That is, if winning alliances are averaging 30 points (very realistic number, IMO), then that 10 point bonus accounts for 1/3rd of their score!

- The amount being scored in autonomous vs. the rest of the match makes the point deficit placed on the autonomous loser quite significant. Both alliances still get the same number of time on offense and defence, so the loser of autonomous has to work extra hard to make up for the points scored in autonomous while keeping pace with what the winners are scoring during the match.

The most significant advantage of winning autonomous, IMO, is being able to defend first. It is very common, early on, for people to not be able to bounce back (as they become edgy and nervous), and for people on the attack to grow increasingly comfortable.

[GaryVoshol]

Originally Posted by M.O.R.T.

I'd say winning autonomous by either means (either winning it by points, or by getting the randomization in the event of a tie) is a HUGE advantage.

Is this not what was asked for? A big criticism of Triple Play was that there was not enough of a benefit for getting the vision tetra.

[David55]

I'm interested in doing such an analysis.
Watching almost the entire 3rd day of VCU, there were many questions that came to my mind about all the different aspects of the game.
The different scoring mechanisms (perfect ball dumper vs. middle level shooter), shooting balls or going back to the ramp, autonomous winning, alliance picking and the outcome (seeing that robots placed 6 and 7 beat robots 2 and 3, and the significance of the robots picked)...etc.

[Greg Marra]

I will create a data collection sheet that can be printed out and taken to Week 2 regionals so we can get the information we need to do a proper analysis. I'll put up both an HTML document for easy printing, and a .xls file for if anyone wants to collect data digitally. All you'll need to do is scan or photograph or type back in the information on the sheet and send it to me and I will compile the master list.

It is interesting that people are saying that when auto was a tie, the team on defense first still won. Once we have a decent set of data to work from, we can start to try to figure out all the variables going on.

[Bill Moore]

Originally Posted by Greg Marra

I will create a data collection sheet that can be printed out and taken to Week 2 regionals so we can get the information we need to do a proper analysis. I'll put up both an HTML document for easy printing, and a .xls file for if anyone wants to collect data digitally. All you'll need to do is scan or photograph or type back in the information on the sheet and send it to me and I will compile the master list.

It is interesting that people are saying that when auto was a tie, the team on defense first still won. Once we have a decent set of data to work from, we can start to try to figure out all the variables going on.

Greg, make sure to include the # of shots taken in your sheet.

Teams that shoot all their balls in auto are obviously aggressive shooters, and may continue to shoot aggressively throughout the match.

I don't care if I only shoot 40% and you shoot 60% if I'm shooting twice as often as you do. I'll win every time just by shooting so much more frequently.

Winning the match may not have anything to do with winning auto, but with the aggressiveness with which the game is played.

[SwitchBlade430]

Even though this is very advantageous, there were a couple of matches in which unstoppable teams like 25 came back and won the match. Congratulations to 25, 103, and 1279 for winning the NJ regional.

[Lil' Lavery]

No, it definately was tied into autonomous mode. The most "Aggressive" shooter at VCU was easily 435, who could unload 5 balls a second and would try to do it 3 or 4 times a match. People quickly caught on and started putting 1 and even 2 defensive bots on them. Because of this, 435 lost 3 matches during qualification, and would lose in the finals (and some, not I, argue should have lost in the semi-finals, because of a semi-controversial DQ). Even when paired with another high caliber shooter, 384, defensive maneuvering and high precision scoring when left open allowed for teh 384, 435, 1522 alliance to be toppled.

This sounds like a really good idea, and I'd be quite interested in getting the data after it's gathered. However, Our team still lost a match after being able to shoot into the center goal during autonomous, and winning it almost every time.

So yeah, Good luck gathering data!