Hi,
My name is Austin and I am one of two of Team 79s Scouting Leads and also in software subteam. I am interested in how other teams do their scouting opposed to our way and see if we need to add anything, or potentially help other teams with their scouting.
I also do pre-scouting in which I do all myself and watch the other teams matches and scout those which we have done for Alabama and I am in the process of doing Houston (Worlds), 50 teams scouted out of the 194 as of now. I noticed some teams pre-scout based on OPR and the data blue alliance site provides, if someone can tell me the advantages to that rather than watching their regional videos and scout that way let me know.
For scouting regionals we take the data that we collect and put it into a spreadsheet and organize and then make a list that Friday night, which I heard most teams don’t do until that Saturday. From Fridays list we only scout the teams that we see potential in and focus mainly on other things such as how their drivers and human players interact and communicate with other teams.
For pre-scouting I have been watching 4-6 matches and take notes on that, if they have more than 1 regional, i usually split it in half so I can see the difference of how good they did between the two regionals. As for which matches I do, I usually do two qualification matches and then one of the Elim matches if they qualified in. We do this to see the difference and try to recognize why they were picked as an alliance or how they got as alliance captain.
Team 79 Krunch this year has decided to not use our usual pen and paper scouting but more of a technical way. I created a form/app that some people call it, although it isn’t an app.
We have used it successfully at Orlando and Alabama.
Works on Androids, Tablets, and Laptops all offline
Downfall to it
Only works on Android, laptops, and Android tablets because Apple likes to restrict a ton of things on their phones’
Not sure if it works on Window phones since those are uncommon and haven’t tested it.
If you are interested in knowing how it works, let me know and I will explain, I also made a github for the code that Team 79 Krunch has allowed me to make public for any team to use.
Again, I would like to know how your team does scouting and structure of it. I think it’s interesting to see how other teams scout.
We used to use electronic scouting but struggled with power supply, data usage, phones dying, distracted students, etc. so we switched back to what I call the old school pen & paper scouting.
We get preliminary info during pit scouting to determine robot “types” (i.e. floor pickup, high goal shooter, etc.). The match scouting determines the effectiveness of those robots.
We make a list Friday night of the top 24 teams (sometimes 24 or 26 if we can’t make up our minds). That list includes people who are higher and lower ranked than us because rankings can change dramatically on Saturday. Then we scout that watch list on Saturday to see if some teams should move up or down on our list. Before our final match on Saturday, we finalize the list and give it to our scouting mentor (me usually) and our student representative. We use paper signs and text messages/phone calls to confirm picks on the field if necessary.
Pros of Pen/Paper
One scouting notebook can be easily viewed and carried around for reference
Simple setup and maintenance
Several team members can review data and specific qualitative notes about teams at the same time
Cons of Pen/Paper
Well, it’s paper in a 3-ring binder so there’s some bulk to it
Difficult to analyze precise data (unless you input it into a database)
Paper isn’t pretty but in some ways it’s simpler to set up and harder to mess up. If you know what you’re looking for in alliance partners, it can be pretty easy to use.
It DEFINITELY help us at Alabama; I think we put together a fantastic alliance (shout out to 79 and 2783!)
We’re using basic paper scouting and have somebody entering the results into a spreadsheet. Our scouts are counting successful gears in auto, gear cycles in tele-op, climb attempts, and successful climbs. They have a few subjective ratings for each team on a 1-5 scale (a 5 being effective at all aspects of the game).
We also pull FMS scoring data (somebody posted a cool script on google sheets that we use) so that we don’t get bogged down in tracking wins, losses, points, # rotors, etc with the paper.
Afterwards we use a data blending tool called Altyerx (FRC teams can request licenses from them), and it runs all the analytics in a few seconds.
Our output report is:
Box plot of scoring for all of the teams (we look for the most reliable high scorers)
Breakdown of teams in 3 buckets (top teams, average teams, below average) ranked by scoring variation, performance ranking (from our scouting sheet), total # gears, etc…
Breakdown of the 10 best gear bots, 10 best fuel bots, 10 best autobots, 10 best climbers
We did have some downfalls because not all of the data got transferred from paper to excel correctly and there may have been mentor error somewhere in the analytics workflow since it was built between 1am and 4am one night…
The paper reports were helpful because the scouts did write notes about teams so when it came to our discussion about who our second pick should be it turned into a really organized conversation when everybody took the reports for 2-3 teams each.
Here are some screenshots from our AppSheet app we used pretty much for pit scouting/pre-scouting: http://imgur.com/a/dkYiB. Regular scouting with this app was pretty sparse, although we had a paper sheet --> app --> analysis workflow that we could have done with more people. Instead, it fell back to a single person (a parent) watching matches and making notes for each team, and another parent put together a short pick list. I didn’t have any time, but I did use Rachel Lim’s component OPR list and provided the top 15 teams in teleop gears, auto gears, and climbing to the main watcher for sanity checks.
Now from limited viewing, it was really challenging to always see what was happening on the field. My very limited data analysis led me to believe that OPR (and most of the components) is only a part of the greater picture (last year I thought it was more predictive, although I haven’t had time to look into that).
To summarize the scouting season (no champs) for me, I didn’t get a Facebook chatbot scouting app that I kind of wanted to try but I did get some practice in hooking up such think using the free IBM Bluemix year trial (worth playing with). The FMS/TBA data was much more accessible thanks to some enterprising people, and I was really impressed with the R package sirirwin made to fetch team data (it is how I populated my pre-scouting spreadsheet and app). There were some nice developments in R with OPR (thanks to Ether), but the data analysis parts for this game haven’t been brought to my attention. I sort of abandoned this way after finding Rachel Lim’s excellent Google Sheet and that Tableau 10 lets you directly pull a google sheet into it. Last year, I used Tableau to display scouting data, including using clustering methods, and with it put together a pick list or more likely a “teams to watch list”. Getting eyes on robots is important, particularly this year.
One of the programmers on my team took the time to create an app for us to use this year, it has a layout of the field and you can interact with the interface to record gear placement (or dropped gears,) shooting or other aspects. And you can record whether they climbed or not, if they got foul points, stuff like that.
We also invested in cheap tablets (Less than $100 I believe each) to use for us to have scouters in the stands and my job was to make sure they did the scouting for each match, and when needed I would scout too (which was alot of the time actually.)
Once we record all the data we put them into Tableau (which sorts data) and we create visualisations and charts to see how good teams did, what they do best and other things in order to determine strategy for matches and for alliance picks when necessary. We also used Microsoft Excel when we wanted to see raw data. The way the app creates the data makes it usable in many sorting tools.
And our team is proud to announce that we have released the app on the Google Play Store, and we hope to create updates for the app when the new competitions for FIRST come out. You can download the app bellow.
What Team 3512 does is we have a form for the 6 scouters which they fill out in the match. After the match those forms are given to the scouting lead who enters all the data into a spreadsheet. All this data is averaged and populated into one sheet where we can see any data we need. This allows for us to make strategy changes quickly. The night after the first day of qualifications we meet in the hotel and make a full ranking list. Then over the course of the second day we ease up on scouting but watch for any robots starting to do better than expected. We have tried electronic forms in the past and they didn’t work as expected, but I’m sure we’d be open to trying it again in the future.
Edit: To organize these papers we used to use a rolling file box, but we replaced it with an accordion folder due to a lack of available space.