PDA

View Full Version : Scouting program in the works, need ideas!


Sean Raia
01-23-2010, 01:35 PM
UPDATED. NEW VERSION, PICTURE POSTED

I have embarked on a quest :yikes: . My quest is to make a simple Breakaway 2010 scouting database program for my team and others who want it.
I figured that before i finalize what data it collects, i should ask all you fine folks here at Chief Delphi for some input.

Here is what i have so far:
http://img714.imageshack.us/img714/5336/scouter.png

Ideas for removing/changing/adding new inputs are welcome!

NOTE:
-each numeric control now goes up to 5 (lowered from 10)
-it will take more than one input for each team and average them into an overall score using a formula.

Chris is me
01-23-2010, 01:46 PM
Way, way more objective data. Subjective data is so much less useful. How many kicks 1 zone forward, 2 zones forward, 0 zones forward into goal, 1 zone forward into goal, 2 zones forward into goal.

How many successful bump climbs, how many failed bump climbs.

Sean Raia
01-23-2010, 01:48 PM
Way, way more objective data. Subjective data is so much less useful. How many kicks 1 zone forward, 2 zones forward, 0 zones forward into goal, 1 zone forward into goal, 2 zones forward into goal.

How many successful bump climbs, how many failed bump climbs.

Okay, you're right. i think that would yield more accurate results being that it isnt based on opinion. Im liking the successful and failed bump climbs idea.

RookieWookiez
01-23-2010, 01:54 PM
Okay, you're right. i think that would yield more accurate results being that it isnt based on opinion. Im liking the successful and failed bump climbs idea.

Probably a little bit on the failing bump climbs is if the team has something on their robot to get them up right again. And if they do, how long it takes might be another factor.

Sean Raia
01-23-2010, 01:56 PM
I like the idea of a checkbox for if they can right themselves if flipped. thanks RookieWookiez

RookieWookiez
01-23-2010, 01:57 PM
Welcome, glad to be of help.

Sean Raia
01-23-2010, 02:08 PM
Come on guys, keep the ideas coming!

RookieWookiez
01-23-2010, 02:11 PM
Another idea might be what's the farthest they can be and get the ball to their own zone.

Sean Raia
01-23-2010, 02:19 PM
You mean like kicker power? i am adding that.

Austin2046
01-23-2010, 02:34 PM
maybe instead of just can they hang, can they hang from other robots? can other robots hang from them?

Bill_B
01-23-2010, 04:01 PM
Program language for your scouter? Labview?

runs on machine - netbook? iphone? regular laptop?

going to need how many programs/machines per match?

BTW - what happened to the wii-mote input devices?

Sean Raia
01-23-2010, 05:06 PM
maybe instead of just can they hang, can they hang from other robots? can other robots hang from them?
To keep it simple, we are just going to have can they hang(meaning from another robot, or the tower). But having a checkbox for if other robots can hang off of them is a possibility. Thanks for the input!

Program language for your scouter? Labview?

runs on machine - netbook? iphone? regular laptop?

going to need how many programs/machines per match?

BTW - what happened to the wii-mote input devices?

1.I am writing the data entry program(the one in the picture) in visual c#, and we are going to have an excel spreadsheet designated for the output (it will come with the program).

2.It will run on any laptop running windows.

3.you could get away with just one, but you can use any number of laptops to speed up the process. I'm thinking of making a file merger that will take all of the data files from the laptops and merge them into one.

4. And i know what you're talking about but i don't know what happened to the wii-mote scouting software.

Phoenix Spud
01-23-2010, 05:45 PM
What about adding an area for comments? This way the scouter can input more information as they see fit.

Will you be making this software publicly available?

apalrd
01-23-2010, 05:52 PM
Idea:
Have a hotkey to add score zone 1,2,3
Buttons for the same thing + flip, hang, allows hanging, etc.
Then calculate ratings when writing to the database based on how many they scored from what zone, and give them a point value with long shots prioritized. Or have it mapped to many game controllers (say 6, 1 per team) to have one person focusing on each robot with only one computer (requiring no network).

Then write a phone app (or something of the like that does not require a full laptop, a pen and notebook (paper not computer) would work), for pit scouting. Have checkboxes for features, and an area for comments. Then feed it into the database and calculate the features score of the robot.

When you get the data, you need to know what kind of robot you want. If your robot can already support other robots, then having another robot to do so has priority 0. If you can't and they can, that might help if you can hang from them.

I began working on such a system, but stopped because I had to develop code for the actual robot. I developed mine in PHP because I could easily network a few of the programming laptops to scout, and it worked nice with MySQL. I used PhpMyAdmin to read the raw data, but never had a working ranking page.

The scouting system we used last year involved pieces of paper with 4 boxes on each side, with check boxes and tally counts for things + comments. We asked the scouts (freshman) to total the tallies at the end, so the data enterers would not have to, but being freshman they didn't. The data enterer would sit somewhere with power and type in all of these sheets, and had a runner to deliver them from the scouts. The data system was an Excel spreadsheet, with linked sheets for Human data entering (what we focused on), ranking, and various calculations. Our robot scouting system was much less technical, involving Chet and Norman writing notes in their notebook for when it came time to come up with the pick list. This system, while labor intensive, worked really well.

Refresh
01-23-2010, 06:08 PM
maybe a button for if they allow for other robots to hook on for the 3 point bonus.

Rizner
01-23-2010, 06:16 PM
How about "How do they hang"

options could be on the top tower, platform, or from a side bar.

Another thing -- would you be setting this up for other teams to use individually, or have you considered having it link together so people could upload for anyone at an event to see?


Also, penalties teams caused?

Bomberofdoom
01-23-2010, 06:55 PM
Before I jump in to the advanced stuff, I think you need to have more control/list boxes for Ball control (including shooting and possesing/herding mechanisim), Drive train (Mecanum, six-wheel drive, SPIRIT etc...), and End-game (Hanging mechanisim, suspension mechanisim (what allows other robots to hang on him)).

In order to store information the information in a database - I say SQL FTW, if you're into it.

Last year, I wrote my own "database" for my (first) scouting application, including serializing the info into byte-size code and saving them into text files which were later deserialized etc...

I'd sugest having two main databases - One is for general information about the team and the robot (usually information gathered in the practice days, about what mechanisims are built into the robot), the second being information about robot preformance in every match.

Eventually, information for tactics for upcoming games need to be given to the drive team very quickly, for the intervals between games are quite short. Having your scouting system having an automated analasyis-of-information system will allow you to give a summarized version of the teams preformance.

But as I see your current application state, I'd say this is an addition you could work on after you've done with the basics.

EDIT: Also for elimination (or even for the quals) - put a "Role" indicator, where you write you analysis about the team's ability to play a certain role in the field (defense, mid-field, scorer, lifter etc..)

PICgnosis
01-23-2010, 07:07 PM
The Inventivity wiimote scouting parameter are summarized in post #54 of this thread: http://www.chiefdelphi.com/forums/showthread.php?t=75784&page=4. Preliminary wiimote button assignments are included as an attachment.

We will also include importation of match data (and possibly images) from www.thebluealliance.net, a robot image database, and printable strategy sheets with team stats and a field diagram.

XaulZan11
01-23-2010, 08:46 PM
[QUOTE=apalrd;905445]
We asked the scouts (freshman) to total the tallies at the end, so the data enterers would not have to, but being freshman they didn't. [QUOTE]

I think of of the most important aspects a scouting system should have is clearity. One thing that I have learned from working with scouts is that errors are usually due to misunderstanding rather than just being lazy. Clear instructions and layout of the paper should be a priority. If it is not, then you will likely get bad data. Bad data is far far worse than no data.

As for the proposed scouting system, I would add something for autonomous mode. Mainly where they start and what did they do.

Chris is me
01-23-2010, 08:49 PM
As a general rule, stuff that's objective that you can get in the pits you don't include in the stand scout sheet. That doesn't mean you take everyone's "we can score 10 balls" seriously in pits and don't check, but you don't need to put whether or not they can hang on the stand scouting sheet, or how far they can physically hit the ball.

apalrd
01-23-2010, 08:55 PM
We once told a pit scout that our robot could fly, just to see if they believed us.

XaulZan11
01-23-2010, 08:59 PM
That doesn't mean you take everyone's "we can score 10 balls" seriously in pits and don't check, but you don't need to put whether or not they can hang on the stand scouting sheet, or how far they can physically hit the ball.

I rarely find value in pit scouting. Other than taking a picture and looking at the drivetrain, I just don't find any value in what teams say about their robot. I'm not saying that every team lies or exaggerates, but I care HOW WELL you can do something not IF you can do it. If you came to our pit in 2007, we would have told you we had ramps. We did, but, I think, we only deployed them twice and never came close to getting anyone elevated. So, a team may have a device to hang, but does it work?

My team hasn't started our scouting discussions yet, but what I will favor is a box that says "Did they hang" and then a general comments section. Thus, we can find out if they score ponts and can write down comments (like, it took them half the match to hang, hung in 5 seconds, tried but ran out of time to hang...) so we know more than just if they did it or not.

CHapstack
01-23-2010, 10:23 PM
I think it's important to include a section about autonomous mode, like how many balls were successfully scored/moved up field in autonomous. You also could consider adding a section that tells which zone a team prefers to start the match in (more pit recon).

davidq2012
01-24-2010, 11:58 AM
I too am working on a scouting program like that. Mine is focused on an easy to use UI and compatibility with XML and The Blue Alliance. Here is a screenshot of what I have for adding a team to the database. It needs some work and I was hoping I could find someone to collaborate with to make a truly effective program.
http://www.chiefdelphi.com/forums/attachment.php?attachmentid=8375&stc=1&d=1264352189

Let me know if you are interested in working with me so we could get more teams to use the same scouting standards.

Sean Raia
01-24-2010, 01:18 PM
What about adding an area for comments? This way the scouter can input more information as they see fit.

Will you be making this software publicly available?

Yes this will be posted on chief delphi when im done. I expect i will be done with it or at least have a beta for you guys in around a week or so... We'll see. And thanks for the idea Pheonix!

Sean Raia
01-24-2010, 01:23 PM
Before I jump in to the advanced stuff, I think you need to have more control/list boxes for Ball control (including shooting and possesing/herding mechanisim), Drive train (Mecanum, six-wheel drive, SPIRIT etc...), and End-game (Hanging mechanisim, suspension mechanisim (what allows other robots to hang on him)).

In order to store information the information in a database - I say SQL FTW, if you're into it.

Last year, I wrote my own "database" for my (first) scouting application, including serializing the info into byte-size code and saving them into text files which were later deserialized etc...

I'd sugest having two main databases - One is for general information about the team and the robot (usually information gathered in the practice days, about what mechanisims are built into the robot), the second being information about robot preformance in every match.

Eventually, information for tactics for upcoming games need to be given to the drive team very quickly, for the intervals between games are quite short. Having your scouting system having an automated analasyis-of-information system will allow you to give a summarized version of the teams preformance.

But as I see your current application state, I'd say this is an addition you could work on after you've done with the basics.

EDIT: Also for elimination (or even for the quals) - put a "Role" indicator, where you write you analysis about the team's ability to play a certain role in the field (defense, mid-field, scorer, lifter etc..)


Thanks for the ideas, i think i will make a small addition that tells you which team is the best and a little about their robot. Although using excel for the read out is possible... im still going to add my own little comparison feature so it can run without excel.

computerteen643
01-24-2010, 02:37 PM
I attached what I came up with for my team.

This, as well as a Pit Scouting program and an access database for both will be available later on for download.

Bomberofdoom
01-24-2010, 02:51 PM
Looks nice, Patrick. I'm starting to see that having a more subjective look on strategy of teams, analyising thier role abilities will be important (Effective blocker, defender, mid-field assister etc....).

Bill_B
01-24-2010, 11:09 PM
Looks pretty good C.Teen. Of course, there are a couple of points:
Take a more careful look at the use of radio buttons versus check boxes. if the answer to the section can contain each of the selections, use checkboxes instead of having a radio that says "all of the above" or "both". let the scout click on all that apply in each case. yes/no deserve to be radio buttons.

The box which records the scout's name deserves to be more complex. Namely, a combo box that allows input and remembers, then lists all the names entered. It will get pretty old for your scout to enter his/her name EVERY time a survey is taken. Let the repeating scouts select a name from the list.

Jeeman
01-24-2010, 11:36 PM
What about type of drive system, mecanum, 6WD, crab, ect? A spot for teams to type in where they are going to play.

Phoenix Spud
01-25-2010, 12:15 AM
Yes this will be posted on chief delphi when im done. I expect i will be done with it or at least have a beta for you guys in around a week or so... We'll see. And thanks for the idea Pheonix!

No problem! I know that our team hasn't decided how we are going to scout yet, but we will need a system since we will be half way around the world without our whole team. :mad: I will bring this up as an idea once we start thinking about how we are going to scout.

computerteen643
01-25-2010, 09:17 AM
the way our scouting system works is fairly complex, a scout records the info onto a custom-made notepad with a form printed on it. This pad is then handed over to a person in our pit, who then logs all of the info into the programs i posted. these programs automatically log and sort the information into an access database for easy access when it comes to the finals.

Sean Raia
01-25-2010, 09:23 AM
No problem! I know that our team hasn't decided how we are going to scout yet, but we will need a system since we will be half way around the world without our whole team. :mad: I will bring this up as an idea once we start thinking about how we are going to scout.
Okay, cool!

the way our scouting system works is fairly complex, a scout records the info onto a custom-made notepad with a form printed on it. This pad is then handed over to a person in our pit, who then logs all of the info into the programs i posted. these programs automatically log and sort the information into an access database for easy access when it comes to the finals.
Your system looks great! What language was that made in, and how long has it taken you to make?

carbuff2228
01-25-2010, 12:56 PM
our team has two sheets. I make two sheets one for the pits and one for the stands. The one for the stands has to be way more complex than the one for the pits. The one for the pit is just basically the team tells you what the robot can do. But the stand one you have to be sure people know how to fill it out

computerteen643
01-25-2010, 01:05 PM
Okay, cool!


Your system looks great! What language was that made in, and how long has it taken you to make?

it is done in visual basic everything including the database will be done in the next couple weeks

Martinez
01-25-2010, 03:36 PM
Might I make a suggestion that may sound a bit controversal?

I suggest adding a line for Pick/No Pick and Why (Offence/Defence/Endgame etc). My experience (over 8 years) is that the hardest thing about Scouting is determing who will work best in your alliance and why. By asking my students what are the particular pros and cons to any potential partner, it gets them thinking in that engineering mind objectively who is the "best" out there but also who might be a "hidden gem." For me in particular it is very difficult to determine a quality second round pick and where I rely the most input from the entire team. By placing such a catagory it really helps shifting through the data at 11:00 PM Sat night.

Also, communicate to your students that Scouting is not just about forming a Pick List. You effectively are the eyes and ears of your team. There is tons of data available but very limited time to gather it in so it is essential to prioritize. Drivers typically only have time to learn about only those teams they have worked/competed with and prepare for the next match, just like a fighter pilot. If your information is quick, compact and well organized then it can provide a solid tactical advantage. My most successful year (Philly 07) with Chuck was primary due to our ablity to feed information from the morning matches directly to the Drivers for the afternoon and beyond.

Its a difficult job that can be highly underated, but if you can explain every team in one to two sentences (weaknesses, strengths and capablities) then that is huge. It will also prove to be a very in demand skill in industry when with the Chief Engineer or major Program Manager and you only has five minutes of his time to get the bottem line.

Martinez
01-25-2010, 03:42 PM
I rarely find value in pit scouting. Other than taking a picture and looking at the drivetrain, I just don't find any value in what teams say about their robot. I'm not saying that every team lies or exaggerates, but I care HOW WELL you can do something not IF you can do it. If you came to our pit in 2007, we would have told you we had ramps. We did, but, I think, we only deployed them twice and never came close to getting anyone elevated. So, a team may have a device to hang, but does it work?

My team hasn't started our scouting discussions yet, but what I will favor is a box that says "Did they hang" and then a general comments section. Thus, we can find out if they score ponts and can write down comments (like, it took them half the match to hang, hung in 5 seconds, tried but ran out of time to hang...) so we know more than just if they did it or not.

QFT. Pit Scouting is what can you do. Field Scouting is how well can you perform. All because they have a kicker, that does not mean it works as well as they had intended or the drivers to pull it off. IMHO it is also important to look at every team over the course of 3-4 matchs in order to get a quality data trend.

Sean Raia
01-25-2010, 05:09 PM
my program is going to focus around field scouting. As far as im concerned nothing tells you how good a robot is or how good their strategy works better than seeing it in action.

seaanenomeenemy
01-26-2010, 09:43 PM
Also put in number of penalties.

Galum
01-26-2010, 10:08 PM
Hmmf here are just a few things that sprang to my mind:

1) Add a part about the drivetrain (e.g omni-directional, can shift gears, pushing strength etc)
2) You need to specify whether the kicker can kick over one bump or two
3) Does the robot have an autonomous mode? Perhaps several? What do they do?
4) How prone are they for fouls? Who commits most of them? Do they carry a yellow card?

Fennig247
02-01-2010, 02:04 PM
Well if this helps out at all. I am the scouting mentor/programmer on Da' Bears and last year I devised a prototype program that I gave to a couple teams to test out. Here is a link of it if you would like to see it. I will be giving it out to more teams hopefully this year if they need help but this might help you get an idea. If anyone needs help with scouting, email me at fennig@me.com. Specially if your doing a vb scouting program, great idea btw.

http://www.chiefdelphi.com/forums/showthread.php?t=76923&highlight=da%27+bears

computerteen643
02-03-2010, 11:57 AM
Well if this helps out at all. I am the scouting mentor/programmer on Da' Bears and last year I devised a prototype program that I gave to a couple teams to test out. Here is a link of it if you would like to see it. I will be giving it out to more teams hopefully this year if they need help but this might help you get an idea. If anyone needs help with scouting, email me at fennig@me.com. Specially if your doing a vb scouting program, great idea btw.

http://www.chiefdelphi.com/forums/showthread.php?t=76923&highlight=da%27+bears

I will be contacting you soon asking for some assistance between vb and access

Sundance Kid
02-13-2010, 01:24 AM
How about something for Autonomous mode? - like if they were able to get all of their balls out of their areas/section.

Also, while a bit subjective, is how well a team plays defense. Since hanging may only be worth up to 3 points, it may be worth it for teams to block a lifter until the last 20 seconds - they may not have enough time to deploy/climb in that time. Or can they block the tunnel, thereby possibly stranding non-climbing bots. This may be worth a lot more than the 3 points of end game.

Let me know if you need any testers or when a copy is available for other teams to try

gblake
02-13-2010, 08:00 PM
Folks,

This is a duplicate of a post I just put into another thread about scouting - If you already saw the other one, you can skip this one.

I'm curious if anyone has tried using the 5th Gear Simulator as a way to test drive their scouting metrics or their scouting recording keeping tools/methods, or their scouts (the humans)?

Imagine this:

Install 5th Gear on enough computers for 6-player matches.
Have 12 of your team members pick a simulated robot and stick with that choice.
Run the 12 players through several 3 vs 3 qualifying matches while the scouts use your team's scouting tools/metrics.
At the end of the quals see if you have successfully identified the best player-plus-simulated-robot combinations and can agree on their rankings and/or suitability for allying with one another.
If you are in the mood, do a draft and see how the resulting alliances fare against one another.

5th Gear stores matches and lets you replay them (in the sense of reviewing them), so if scouts are missing important observations, replays can be used to train them; or if scouts are disagreeing about how to evaluate a robot/driver's performance they can look at the replay together and try to reach a consensus.

Finally, while I think this can be a great way to evaluate scouts, metrics and tools, a word of caution is apropos. Remember that a simulation is not reality. At best it is an useful approximation of important aspects of a match and leaves out others. For example, 5th Gear doesn't simulate what happens when a harried pit crew forgets to install a fully charged battery, or when a weak axle or weld bend or breaks under stress during an important match, or ...

Blake
PS: 5th Gear doesn't simulate end-of-game hanging/lifting. To factor this into a scouting exercise; roll dice to see which robots that were in position at the end of a match should be given the bonus points (while the other bots continue to score regular points).
PPS: 5th Gear doesn't assess penalties. Assign a few people to referee duty and let them keep track of penalties if you want that level of realism. They can watch over the shoulders of the players and/o watch the server's monitor.