Scouting Sheets

[drhall]

Hello,

Looking into redoing our scouting sheet

Any thoughts on what areas should be on a scounting sheet.

Thank you for any input you make have.

Dan

[bigbeezy]

i know we have in the past looked at mainly numbers. Like number of balls scored/attempted, HP balls scored/attempted, EC's ran to HP, SC's Scored. In previous games we'd look more at autonomous routes.

This is my first year in the stands, but what I've been looking at is speed and consistency. You could be a really fast scorer but if nearly all the balls end up on the ground that’s not good, and also a very accurate robot that is slow is not so good either; differences raw numbers sometimes fail to pick up. I would make a note of teams that failed to move during auton, or entire match. Look for jams and failed mechanisms. Then I'd try to the best of my ability unbiasly rank the teams based on speed for different parts (i.e. a shooter/dumper was fast, but their loader was slow). Then i could use this data with the others that have more of the raw data to help make a pick list.

[JHay]

Also, you should have their actual autonomous patterns written down. This could help you coordinate strategies if your on their alliance, or find out how to counteract them if they are on another alliance.

Although it isn't relevant in most cases, starting positions are good to track where teams do good and bad when they go into autonomous. (i.e. A robot that goes straight in autonomous would not do so well if in the center. They would just run right into the other side...hard.)

I agree with bigbeezy that the success rate for the amount of moon rocks made into a trailer will help you determine your better offensive robots.

Defense capability should also be marked down as it is now becoming increasingly common (at least in FiM) that many teams need someone to help *cough*:rolleyes: "nudge" opposing robots away.

[NorviewsVeteran]

I'd leave lots of room for notes and comments.











Maybe that much room.

I also read something about an unmarked check box at the bottom of the page called the 'jerk box'. Most teams want to keep that box unchecked.

[Brad Voracek]

The most important statistic, by far this year, is how many balls are in the teams trailer at the end of a match.

If you want to go by pure numbers, that's my favorite number to look at. Then it goes on to speed, scoring capability, and ball pick up ability. All of which are qualitative rather than quantitative.


Another thing, this year, wide robots tend to be more maneuverable, so they automatically get a bonus in my book. Along with swerve drives.


If opponents can't score on your teams trailers, you win. Easy as that.

[smurfgirl]

It's important to leave room for both qualitative and quantitative analysis. While it's good to know that Team A has 85% in scoring, it's also good to know that you can demolish them by bumping them in the rear right corner. I'm going to link you to what one coach likes to see from her scouters (http://uberbots.org/forums/viewtopic.php?f=37&t=423), though I'm sure there are many other possibilities that are equally as good if not better. Hopefully you'll at least get some ideas from that. (Sorry for the outside link, but it was a lot to paste in.)

Edit: Also, I'm not sure if this is something you can specifically scout for from the stands, but by interacting with teams through pit scouting, and talking to your coach who has worked with teams on the field, please check up on teamwork. If a team has a great robot, but is unable or unwilling to work with other teams, achieving success with them will be very difficult. Look carefully at how they interact with others, and if they are willing to compromise to bring success to the entire alliance.

[JDM]

Team 2199 keeps a very simple scouting sheet, which allows us to recall specific traits about each robot to go along with team members' general observations. It works well for us, but if you don't have several people who stay generally aware of the teams and how their robots are doing then it might not be best. If you have lots of scouts it might be better for you to go with a complex system that simply gathers lots of data.

For pit scouting, there's a series of yes/no questions that reflect traits we would look for in alliance partners, a ranking for team attitude, and a large space for comments. Simple questions like "do you have an autonomous mode" narrow the amount of teams that we consider quite easily, and team attitude also plays an important part of it.

As for field scouting, go with whatever you think is important to you. Our sheets had 1-5 rankings with general categories reflecting the robot and HP's skills in maneuvering and delivering cells, as space for win/lose, penalties, and comments. It was then entered into a spreadsheet later for analysis.

Since I'm talking about it here, I really have to take the time to thank our head scout who came up with this system once again. Without this, we couldn't have done so well.

[MrForbes]

We made up pretty simple scouting sheets this year

[Phyrxes]

We used a largely paper based scouting system this year it will probably migrate slowly to computer based but paper and clipboards I can safely hand to freshman.

Our pit scouting sheet was a series of basic check the item that is appropriate for this robot.

We kept human player and robot data separate to make it easier to organize. For human players we looked at penalties, shooting percentage, range, and super cells. The robot data was more complex so we could get a better picture of the robot/drive team as a whole not just does it score a lot and does it get scored on a lot?

[roboraven15]

definatly you cannot forget how many balls are in your trailer at end of match. its a stat not always looked at, like a team at the last competition was a good scoring robot puting lets say 15 balls into trailer each match. only problem with them, they averaged prolly 20+ in there own.

[rogerlsmith]

We are looking a lot at scoring averages. A quick way to get this info is to use the "Lunacy Scout" at http://www.firstobjective.org. Type in any team number for a complete analysis. We are putting this info right on our scouting sheets before the matches begin.

[prettycolors91]

One of the most important features of our sheets is the mini-field diagram. Scouters draw the course of the robot during autonomous, and over the course of the regional, the strategists/drive team can discern a pattern in a team has multiple autonomous programs. We also assign about 4 kids to a robot, which allows them to count how many shot as opposed to scored, the accuracy of the human player, speed, amount scored on, etc.

[martin417]

Scouters also need to talk to the drive team after every match. What is important is, how well did each team work with your team? did they listen to your suggestions? Did they do what you asked of them? Did they ask you to do something that did/didn't fit well with your bot/strategy?

The best team in the world is useless if it doesn't fit into your strategy or meld with your strengths.

[Anders Horn]

I ran the numbers and it doesn't look like #of balls in your trailer has any predictive power at all.

[MrForbes]

That's interesting....even if you're a no-show?

[Anders Horn]

I don't know the explanation for it but a few days ago I decided to test how well my scouting sheets worked, so I used various measures to try to forecast match results and then compared them to the actual matches.

When I made a graph where x was expected match score [based on average #of points scored for each robot] and y was actual match score [each match was an x,y point] I got a fuzzy oval inclined at 45 degrees [it seemed to me to be slightly less than twice as long as it was wide].

When I tried the same thing with the average # of points in each robot's trailer I got a perfect Gaussian circle centered on the median. I then constructed a perfect Gaussian circle just to check. And so I came to the conclusion that statistically #of points scored on you was not a useful prediction.

It may be a useful predictive factor for a few select robots, but I don't have to data to know.

[MrForbes]

Very interesting! Thanks for the info....

[Anders Horn]

http://kamocat.com/scouting/moredata.html has pictures of the graphs if you want to analyze them for yourself.