How accurate does scouting need to be?

[wesleyac]

Hi,

I'm part of the programming team that's working on the scouting system. Our team is trying something new for the scouting this year: Tracking position of the robot as well as actions. (Among other things) From our tests, it seems to me as though this will make the scouting less accurate.

From doing this, using last year's scouting system a bit, and doing paper scouting, it seems like there is a tradeoff between accuracy and amount of data.

So, my question is this: Would you prefer to have lots of less accurate data, or not as much accurate data?
(This isn't going to affect our scouting system, I'm just curious.)

[George1902]

Inaccurate data is not data. It's guessing.

Collect as much data as you can according to your resources. Even if you can only tell part of the story, it's better than fiction.

[Qbot2640]

The following is anecdotal and opinion based - so I encourage others to chime in in support or disagreement - but I believe timeliness of data is going to be paramount this year. With only one game piece in play per alliance at a time, knowing your alliances strengths and weaknesses as well as those of your opposing alliance for each match will be critical. Good data available in a timely manner will enable the proper strategy discussions to take place that will win matches. Strategy and full alliance implementation of that strategy will beat technological prowess this year.

Also - the best teams will play different matches with a completely different strategy - all based on the "mix of six" on the field.

[matthew_martin]

As one of the scouting leads on my team (pit scouting and match scouting), I say that inaccurate data is literally as good as a robot that is just a chassis. it might get a little done, but it'll never make your scouting system fantastic. Plus, this year, scouting is of the utmost importance. It's a good idea to meet with your alliance members BEFORE your match to discuss strategy and work out the game plan. I've always said that scouting is like fantasy football. You can never build a winning alliance without good data.

[rsegrest]

Insert programming adage here...Garbage In Garbage Out...

If your data is garbage or extraneous then so is your information.

George is correct. Collect as much data as you can that you can make sense of.

In our scouting what will be important are starting position, assists, truss points, catches, and goal points (not necessarily in that order).

[EricLeifermann]

Originally Posted by wesleyac

Hi,

I'm part of the programming team that's working on the scouting system. Our team is trying something new for the scouting this year: Tracking position of the robot as well as actions. (Among other things) From our tests, it seems to me as though this will make the scouting less accurate.

From doing this, using last year's scouting system a bit, and doing paper scouting, it seems like there is a tradeoff between accuracy and amount of data.

So, my question is this: Would you prefer to have lots of less accurate data, or not as much accurate data?
(This isn't going to affect our scouting system, I'm just curious.)

The last 2 years we used a tablet program that tracked what teams did and where they did them on the field. I can say that we never used data gathered on where they did what on the field. Collecting that data just slowed my scouters down and increased mistakes.

This year we are still trying to use a scouting app but we will not be and do not care about what specific location a team does something. I suspect our scouting to be much better and allow us to pick those robots that will really complete an alliance in the 2nd round....

[Tuba4]

I submit that shot location information can be very important this year. If a bot has a small distance window or can only hit from a specific spot or spots that info makes it much easier to defend. A simple x on a field diagram marks a shot. Circle it if made. It can be that simple. But watch out for the bot that can shoot but shoots very little. That is the one that in the third match of the finals surprises everyone by rolling out to thirty feet and nails one in the high goal.

[wesleyac]

First, I'd like to thank everyone for the replys.
It seems to me as though I may have misrepresented our team's system. To me,it seems that all scouting data is flawed, just due to the fact that humans are doing the scouting. Our data won't be much worse than paper scouting or tablet scouting without position data, and with scout training, the accuracy issue might be negated.

To me, it seems like Garbage in, Garbage out is a scouting issue, not a tablet scouting or position tracking issue. Having been a scout, I know that all of the scouts will make mistakes at some point.

Having said that, I agree that we should collect as much data as we can according to our resources. We'll see if that includes location data or not.

[Animal Control]

Ask james@team2168.org he has a scouting sheet.

[Abhishek R]

Tracking position of where teams liked to shoot from last year was useful so you didn't have to share a spot, but eventually most teams was able to shoot from the corners and the back of the pyramid, or FCS. However, this knowledge could've been gained from pit scouting.

This year it may be to your advantage to figure out what spots/zones teams like to play the most, as this game is likely going to evolve into position based play. With only one game piece for the entire alliance, scouts shouldn't have too much trouble keeping track of positions as well as other data, however, as you already know, there will always be some mistakes made.

[Team 2016]

Quality over quantity. More accurate data, but less of it is better. Tons of data means nothing if none of it is accurate.

[otherguy]

Originally Posted by Animal Control

Ask james@team2168.org he has a scouting sheet.

Since that me, and I've gotten e-mails referring to this thread.

I can honestly say that I do not.

1. I don't handle scouting for our team
2. Our team hasn't settled on how we are scouting yet

I will use this opportunity to plug one of our past students' android apps that afaik he still maintains: https://play.google.com/store/apps/d...RC_Scout&hl=en

It's a customizable scouting system. You choose what data you want to collect and everyone on your team can scout matches from their phones.

[Citrus Dad]

Originally Posted by wesleyac

First, I'd like to thank everyone for the replys.
It seems to me as though I may have misrepresented our team's system. To me,it seems that all scouting data is flawed, just due to the fact that humans are doing the scouting. Our data won't be much worse than paper scouting or tablet scouting without position data, and with scout training, the accuracy issue might be negated.

To me, it seems like Garbage in, Garbage out is a scouting issue, not a tablet scouting or position tracking issue. Having been a scout, I know that all of the scouts will make mistakes at some point.

Having said that, I agree that we should collect as much data as we can according to our resources. We'll see if that includes location data or not.

Just to follow up on Wesley, 1678's tablet based scouting system appeared to be quite accurate last year. Our average match scores and the correlation with the OPR were quite high. We were extremely happy with the results. We also were able to get near real time data to our drive team electronically.

This year is a bigger challenge because of the direct linkage among alliance robots and that there's no single position on the field that might be an attractive spot, i.e., safe zone or structure. Positioning for the truss shot is another added dimension. Our scouting system is much more complex this year than year for this reason.