I am curious to hear about other solution to a ranking algorithm. Although I don’t think you can completely rely on quantitative data in this game, it can defiantly be helpful.
My team ignored fuel for week 1, and tacked how many teleop gears a robot would score, if they would score a gear in auto and if they would climb
Now where it gets tricky is the playoffs, particularly the 2nd robot you would pick (assuming you are the alliance captain) and how to rank a robots value.
50 Points regardless is great, but the trend seems that generally the top 24 robots at a competition are very capable of climbing somewhat consistently
I am currious to hear how teams are ranking each, I guess teleop gears are the most important but how do I decide what the better team is, the robot that can contribute 4 gears in tele op, or the one that does 3 and can climb.
I guess it would matter on if you think you are going to be able to get 4 rotors!
Auto gears are also really important…Would love to hear what people think!
I think the way to look at that scenario is how sure are you that you will get the 140 point bonus (in lieu of a nearly certain 50 point climb) ? If you fail at 140 thats a lot of wasted time, might as well take the 50.
I think it comes down to how prolific the gear bots you assemble are, do you have some fuel capability for tiebreaker? Are there likely very good defenders lurking?
I think it really depends on the teams at each event. So never a standard answer.
I was watching Southern Cross, alliance #1 was at the 30 marker three gears from 140, they climbed. I thought they should have got the 4 rotors , they certainly could have got 3 more in last 30 almost guaranteed. I suspect they knew climbs would win it,…and did. But just shows 4 rotors is possible and that trumps two climbs. 100 vs 140
We typically try to be brutally honest with ourselves what we can do (scout us just like any other team) , the fill in the performance voids with the alliance partners, we always approach the event with a captain mindset and look for two that help us build a winning strength alliance. Luckily teams ahead of us typically think the same way. When we are the captains first pick our all event deep scouting usually help the captain pick a third we already identified compliments us. In the end three solid picks is what wins these events. If there has ever been a qualitative scouting event this is it.
As for quantitative… it goes Gears, climbs and the rest typically as to who gets picked from what I see at events, with the top prolific auto Fuel bots using ranking high enough to be a top captain. Probably not going to find a lot of Fuel help after the captains anyhow.
I would encourage you to split robots into categories based on what roles they can play, and rank them on their ability in that role. Teams can go in any number of lists. If they do it all better than anyone else, they can go at the top of every list.
You could have a seperate list for fuel, gears, gears and defense, gears and fuel, whatever you want. If you want a gear bot and a gear/defense bot, pick the best team you can get from each list. PLUS you can quickly reference what other alliance are building for their composition.
The specifics of how you split and rank teams are up to your team, this is purely an organizational method.
Yeah… your answer is pretty much what I figured was the tradeoff, deciding to shoot for it all gear wise or going safer on climbing… our strategy week 1 was looking at our first pick, and seeing what we could do gear wise and then trying to determine if the 4 rotors would be possible.
I can’t speak loads for this year, due to our not being in a picking position thus far. We knew it, so we set about making our “helper list” with different methods. For as long as I can remember we have run out scouting as a sort of discussion, and less of a purely numerical approach. Once we can sort teams into rough “pools” of good v.s meh in shooting, climbing, etc with scouting data in depth discussion follows. It is a very organic process. Occasionally, with proper preparation and a clear alliance goal, we can get these meetings down to ~45 min.
The numerical data we gather is primarily applied to individual qualification match reports. There are only about 3-5 numbers we look at for generating a picklist. As far as any sorting algorithms for ranking: I have found 2015’s “qualification average” to be a relatively telling of where a team falls in the stack and is dead simple to integrate into your system. This year we didn’t bother trying to calculate a team’s individual contribution as it is so dependent on your alliance partners and rather difficult as fuel becomes more prevalent.
Reaching rotor thresholds complicates scouting this year. You have to first determine your team’s strategy, e.g., 4 rotors vs 3 rotors and a more likely climb, and then scout to that. You can’t do it the other way around.
If you want more general guidance, you can find our scouting whitepaper on our 2016 system (which is very similar to this year’s) on CD.