Oh, boy, scouting will be difficult this year!

[jayjaywalker3]

Originally Posted by kirtar

During a break I jotted down any item that I could reasonably want to keep track of, and I came up with too many things to keep track of simultaneously (e.g. not only how much does the robot score, but how much is it scored on?). The freshmen... mean volunteers aren't going to like this.

What did you write down?

[kirtar]

Here are a few for the robot: Moon rocks scored by robot, empty cells transferred, moon rocks scored on robot, penalties, various comments, approximate hit percent (not happening, and even then not going to be a big factor), score differential (this is more likely to be calculated by total team score), super cells scored on/by robot (probably too hard to spot at distance).

For the human player:
#In during autonomous, #miss during autonomous, #In during Tele, #miss during tele, Fueling station/outpost, Hit percent (this will be calculated by access from previous data), Penalties (Hopefully this will always be 0), comment, super cells thrown out of # available (this is going to be HARD).

Since there will probably be robots that just dump moon rocks in, it may be very difficult to count how many go in. Also, teams may vary the number of moon rocks that are preloaded, although I predict that most will go with 7 unless that cannot hold that many. At this point I'm trying to focus on quantitative data, and there's plenty to keep track of that could be useful but will likely be inaccurate.

I chose to split autonomous mode and teleoperated mode for the human player because the robots start in front of an opposing player so there is at least a brief window where the robot should be a relatively easy target with the low acceleration that is possible on the field. Also, during autonomous mode, the robots cannot actively avoid moon rocks, although they will likely move in a path that is not meant to be easy to hit (if the robot is not tracking on to a vision target).

[GillSt.Bernards]

that is a good idea. That will help keep organized also last year my team made a really simple robot. The judges really liked it. Maybe this year simpler is better. it si something to keep an eye out during scouting.

[dividedby0]

I agree scouting will be hard this year. There are a lot more important roles in the game this year. Shooters and Robots alike need to be looked at. I hope as my team's scouter it won't be to hard. Good Luck all you scouters!

[Lkr220]

Maybe because we already use Wifi for programming, there should be some push to, at each regional, get a "server" running on a public WiFi where people can post up the data they recorded (as well as access this data)?

Just an idea.

Team 1676 has done this for the past 2 years at every regional we attend. So if you go to the NJ or Palmetto Regional this year you will have access to our scouting data if you have a laptop.

Since we use our school laptops to scout (instead of the traditional paper and pencil) the database is always up-to-date.

[smurfgirl]

I think scouting will be difficult, but manageable.

A suggestion is to begin by taking pictures of all the robots on Thursday, and filing them ahead of time so you can remember a robot when you look at its file. Then, perhaps use a system with 6 scouters, 1 per team on the field. Have them keep track of key elements:

- how they pick up
- how they score
- which pieces they score
- how accurately they score them
- how many they score
- their driving ability
- their defensive capabilities
- any penalties they incurred
- their weaknesses
- how easily they can be scored on
- where their PS is positioned
- what their PS does during the match
- what they do in autonomous
- if/how they interact with other robots
- do they get stuck in robot pile-ups?
- do they break down easily?
- etc, I'm sure there are other useful things that I forgot to add

While this sounds like a long list, it's actually relatively easy to get all this data. Before the match starts, check on PS positioning, and then just watch the robot and write for the duration of the match. Perhaps something like:

"PS in home fueling station, auton. drives straight to other side of field, gets moonrocks from PS, shoots into other robots, ~50% accuracy, scored about 8 moonrocks on other alliance, lost control of robot and tipped about 1:20 into match- very vulnerable when hit sideways"

You can scrawl something like that out over the course of the match, and since you'll have data like this for every match, you can get a pretty clear picture of how the robot performs, and its trends for the day, rather than looking at one isolated match (the last one it played on Saturday, because you remember it as you're scrambling to come up with picks for elims), or raw numbers (their match scores were 30-50, etc.)

If you keep track of this by computer and use google docs or the like, you can combine this into one spreadsheet, or if you do it on paper, you can organize it numerically by team, and each scouter can put their pages in the scouting binder in the proper location. You'll have a wealth of useful data by the time you need to start thinking about alliances.


Edit: Oops I forgot the other thing I wanted to say. You can join up with other teams and share scouting data, creating databases to share, or you can have your scouters work together if you don't have 6 people to spare. (: Lots of teams do this, and it works out great!

[Ian Curtis]

I'd highly recommend some sort of quantitative system (i.e. keeping track of balls scored by Team XXXX) vs. a qualitative system (i.e. Team XXXX scored a bunch of balls). That way you always have an objective observer, and people's personal bench marks for "good", "bad", "awesome" don't interfere with your data. Not to mention, it makes it way easier to plan match strategy, both in terms of your opposition and your partners.

Everyone overstates the capabilities of their robot. We don't do it intentionally, there's just a lot of pride and worked wrapped up in these magnificent machines. So, it's really helpful to your alliance when you know that team XXXX has actually been averaging Y orbit balls, instead of the Z they told you. Also, it's really cool to see how you stack up on Friday night.

That said, handwritten notes really round out your data. Not everything is obvious in the averages. These handwritten notes can push you to examine spectacular teams you may have missed otherwise.

Also cell phones. I live in the dark ages, and have yet to purchase one. But on Friday/Saturday, they are a total lifesaver. I can't carry around all of our scouting data in my backpack. But the scouting team in the stands can look anything up while we wait for our match to start. It's really, really nice.

[bduddy]

Originally Posted by iCurtis

I'd highly recommend some sort of quantitative system (i.e. keeping track of balls scored by Team XXXX) vs. a qualitative system (i.e. Team XXXX scored a bunch of balls). That way you always have an objective observer, and people's personal bench marks for "good", "bad", "awesome" don't interfere with your data. Not to mention, it makes it way easier to plan match strategy, both in terms of your opposition and your partners.

Everyone overstates the capabilities of their robot. We don't do it intentionally, there's just a lot of pride and worked wrapped up in these magnificent machines. So, it's really helpful to your alliance when you know that team XXXX has actually been averaging Y orbit balls, instead of the Z they told you. Also, it's really cool to see how you stack up on Friday night.

That said, handwritten notes really round out your data. Not everything is obvious in the averages. These handwritten notes can push you to examine spectacular teams you may have missed otherwise.

Also cell phones. I live in the dark ages, and have yet to purchase one. But on Friday/Saturday, they are a total lifesaver. I can't carry around all of our scouting data in my backpack. But the scouting team in the stands can look anything up while we wait for our match to start. It's really, really nice.

For the years I did scouting for my team, I made sure to focus mostly on numbers and data for the one-team sheets, although I did have a "rank-out-of-5" section. Last year, though, I printed out a sheet (double-sided) with every team name and number next to it, so I could keep it on a clipboard and write little comments myself. We weren't anywhere close to picking, of course, but it would have helped, especially considering that I forgot alliance selection was during lunch, not after

[shannieliz]

My opinion on the human player is that as long as a team has somebody who can aim AND predict the next move of the drive teams, they will do fine and will score a good amount of points. A basketball player in particular would be almost perfect as a human player. If you think about it, in a basketball game, they have to have good aim to get it to their team mates and also predict what the other team is gonna do to attempt to block them. Baseball and softball players might work, but not necessarily.

[1337Nerd]

I always avoid scouting anyway.

I'm the only one on the team who can film properly without the camera flying all over the place.

Grade 9's are useful in the scouting department.

But I'd say scouting will probably be like every other year and just concentrate on the robot. If you happen to notice that their human player is good, more power to you.

[kirtar]

Originally Posted by 1337Nerd

I always avoid scouting anyway.

I'm the only one on the team who can film properly without the camera flying all over the place.

Grade 9's are useful in the scouting department.

But I'd say scouting will probably be like every other year and just concentrate on the robot. If you happen to notice that their human player is good, more power to you.

Yeah they're useful... until they throw their clipboards on the ground and leave to go to the CNN center early (this happened last year). I started last year and somewhat ran scouting from a back seat. After about the first day at the Championship I just gave up since there was no way that we would be picking and highly unlikely that we would be picked (e.g. if I was picking, I would not have picked my team because of my data). I'll be looking at human player separately, but if it doesn't seem worth it after (or even during) Buckeye I'll scrap the HP scouting completely in favor of notes on good HPs.

Also, random fact: After divisions for the Championship were announced last year, I told myself, "If 1114, 217, and 148 get on an alliance, they're probably going to win. Not like that's going to happen, since one of them will get picked before they can form that alliance."

[JDeCola]

Originally Posted by GillSt.Bernards

Get some of the baseball, softball, and basketball players to join the team. To play those sports you need to have good aim

throwing an orbit ball is a lot harder than throwing a basketball or baseball, they would need a lot of practice, something we are all scrambling to get rightnow as far as the human player in the game

[Nibbles]

Originally Posted by smurfgirl

- how they pick up
- how they score
...

While this sounds like a long list, it's actually relatively easy to get all this data. Before the match starts, check on PS positioning, and then just watch the robot and write for the duration of the match.

You only need three items from that list. Separate design elements from success elements. Accuracy is important, but not for success. Of the ones important to keep track of, only these need to be watched out for during the game:

• Game pieces scored (Sheer numbers - score contributed - is all that matters economically)
• Game pieces in trailer (Might be useful)
• Shortest time from capture to score (If you need to score super cells)
• Activity in autonomous mode (The robot isn't a sitting duck, or, the robot is efficiently designed)

The first two you can get from counting alone, the third takes a bit of studying and you may want one person dedicated to this. The second one, game pieces in trailer, may not even be necessary, and the fourth one, autonomous mode, could be inferred from either the score at autonomous or the second point. Usually I have been able to scout with just two people, this year may require three for me though. Penalties are also important, though you can figure those out after the match (maybe).

Too often people think that there is something significant in what kind of drive system the robot has, say, when it just doesn't matter. There are several different areas a team can be strong in, and when picking alliances you want to be well rounded in all. Things like has a camera simply doesn't affect any of them (by itself). In the last few seasons there have been two distinct statistical areas, teleop and autonomous (and each of those typically can be broken down into about two subareas), and I am predicting this season there will be only one major one, and a minor human player element which will have a deviation too large to be significant. Those two areas (or this year, maybe, one) are all scouters need to be concerned with.

I come to these conclusions by an economic analysts. Start with the theory that resources are limited, and consider the arena a system that contains economic goods, game pieces, goals, and time. The idea is to identify all the resources used to score points and determine how their cost changes with different robots.

This season, we have a few economic goods on the field: The corners, each of the game pieces, trailer space. Cameras, number of wheels, shooter type, ect, has no bearing on how these resources are utilized (well, it does affect how, but you cannot calculate how, and it isn't what is important in making a decision anyways). What does matter is numbers: the cost (resources used, including time) per point. When doing scouting, look for total score contributed by each team, and what other teams have to give up to score. This year, there is not very many things one team has to give up to allow another to score a point, with the exception of super cells for empty cells and moon rocks, since they have a different point value. Autonomous and teleoperated game pieces are scored equally, so there is no time preference on them with respect to the score. If there was, like last year, robots that can do laps vs. score over the overpass would be a consideration. In short, there are very few parameters to be concerned with when selecting alliances, the only major one being how many game pieces can you score (what is the price in time), the other insignificant ones, retrieving empty cells (what is the price per empty cell exchanged), and how well can the human player shoot (what is the price in attempted game pieces, I wouldn't even consider this significant enough to record). All you need to know during scouting is data related to these, which is (as is usual) very easy.

You may be surprised you don't need as much data as you think you do.

[Jimmy Cao]

Scouting will take more more resources, and more "training" this year.

Last year, you could rather easily count the # of balls a team scored, and that would be a decent representation of that team's offensive ability. Granted you can still do that this year, it would be 1) more difficult to count, and 2) (IMO) a less accurate representation of the team as a whole. I would think that it would be useful to somehow judge how easy teams get scored on. Ex, if a robot scores 20 balls/match, but always manages to get its own trailer filled, then maybe a robot that scores 15/match but manages to only pick up 1 or 2 balls/match would be a more helpful partner.

Another thing that can play a more significant role this year is Human Player ability. That might be even harder to notice.

Most importantly, however, is that the scouts need to know what they're looking for, and as long as you have a few people getting good, useful data, then I think scouting won't be an issue.

[kirtar]

Of course. Last year the teams that were picked (in the competitions I went to) scored rather consistently every time with some anomalies (e.g. 71's arm being broken in the middle of a match), and thus by even getting one match for each team it could be possible to get an idea of who was good and who wasn't. However, this year, the scoring is highly dependent on your opponents since they are your targets (well actually, the trailers behind them, but they control the motion of said trailers).

Of course, raw score is not as important as what I call "score differential." As you mentioned, it is important that you score more than you are scored on. In fact, I'll probably even ignore scored supercells and focus on the total number of balls that are scored. This year, data will be less important and for the most part, I'll let the human element always override the data, with the data providing a very rough idea. With all of the variables (i.e. scoring on your opponents), it will become more important to judge by watching rather than by running averages.