I think I have sufficient data to do this analysis. Will spend some time on it tonight
Yes, we did realize that. But, no, it will not be hard at all and still done with two lines of code 1) motor to 100%. 2) drive forward X inches.
Without having looked at the data, I will say that, for a first year with AWP, I think Change Up’s AWP criteria were excellent. High-level teams could do it on their own, with sufficient time and dedication. Lower-performing alliances could achieve it, though it required at least one of the robots to do slightly more than the “minimally viable scoring robot”.
This year’s, on surface, seems…underwhelming. I do hope that the potential strategic choices make up for how seemingly simplistic it is, as well as the low barrier encouraging teams to write minimally viable autons.
Agree. I always thought that the Autonomous Bonus was there to encourage teams to write auton. Getting a WP (1/2 a win) with just a push bot and two lines of code seems excessive to me.
I’ve excluded Leagues from this.
Based on this, maybe an ideal state would top off somewhere in the 30-50% success rate? In a tournament with 8 qualifiers, teams earning an extra 3-4 WP seems good (maybe bordering on generous). Enough to mix up the standings a little bit over direct WLT records. That rate seems like it would preclude pushbots from succeeding thru direct contribution though…
Tough balancing act for the GDC. I wonder what their target is…
Two AWP could be earned in a match, correct? so for 2020-08:
|Year Month||QualificationMatchesPlayed||AWP Possible||AWP Earned||AWP SuccessRate|
Either way, all the numbers look pretty low…
Yeah, I maybe should have explained the numbers a bit.
QualificationMatches = Sum of WLT for all teams at the tournament
AWP = Sum of the inferred AWP (e.g. WP- (2W + 1T)) for all teams in the tournament
Last year was, obviously, an odd year. Many teams didn’t play until at least Jan-Feb, but that is some good data. Maybe last year’s was a bit hard, but not by much.
This year will be A LOT higher.
I think Change Up’s was pretty much goldilocks. From a “it takes 2 ‘minimally viable’ robots to perform” perspective:
- One robot needed to:
- Move towards a goal (since it realistically couldn’t start positioned to score directly into a corner goal)
- Shoot a ball into the goal
- The other robot needed to do at least what the first one did and:
- Get a second ball
- Move to a second goal
Obviously highly skilled teams were able to perform these actions on their own, with some amount of risk of failing.
Interesting data. What is included in the April and May numbers? There were a lot of LRT scrimmages in that timeframe, which were not as competitive. Is Worlds included in the May numbers, 370 seems low for the match count.
The AWP for this year is way too easy. Hopefully, someone from the GDC is reviewing this thread and will consider a change in the next game rules update.
Showcase events were in may.
Analysis only included In Person Tournaments, so no LRTs (and to be clear, Worlds as an LRT is not included)
Would be interesting to see the numbers for LRTs as well, now that the results for a lot of the LRT events (maybe all? I have only checked a few) are available on RobotEvents – although I’m not sure if it would super relevant to the discussion in this thread.
My guess would be that the auton WP was significantly more common at LRT events, because (a) you didn’t have to worry about getting in your partner’s way, (b) you could own a goal in your home row in auton just by wedging a ball in the bottom, and maybe (c) teams participating in LRTs were more likely to have more complex autonomous routines.
Perhaps; the LRT game was significantly different than in-person last year. This year’s is even moreso. where teams may earn up to 6 WP per match.
The data from RobotEvents API made it a bit difficult to differentiate Change Up’s LRT from InPerson events, and LRT events weren’t even in there until late in the season that I found it better to exclude them entirely.
I may be a bit late, but this right here. This right here is the absolute best thing that could be said about the AWP for this season. Thank you for posting.
As it has been stated over and over again in this thread, the AWP is meant to be a means of incetivizing all teams to do an autonomous routine. What a terrbile way to incentivize something if it took 12 hours and 60 lines of code. Especially if you had written 0 lines of code before this.
If it took your kids 20 minutes to accomplish that task, I’d love to see what they could do with a polished robot and an entire season to code an auto that does more than just score a single ring and move a single goal.
Agree to disagree. I love the idea to incentivize programming autonomous. I just don’t think two lines of code 1) set motors to 100% and 2) go forward is auton. You shouldn’t get 1/2 a win for that. It will now turn into a penalty if you don’t get it somehow - making the oddness of scheduling even more so.
There was more 10x more work out into building that robot than programming. Why is it wrong to incentivize a few hours of programming? The whole game requires hours of building - why the drastic unequal balance?
I think you underestimate how many people don’t write code. At the Kansas middle state comp not 1 team had an autonomous.
to be fair there are other rewards for a more complicated and competitive auton as well as the win point reward for a simple one. The win point will help you in your rankings, but getting neutral goals from the center into your home zone is both highly advantageous strategy-wise, and worth a lot of points towards winning the auton bonus, which is significant.
I think the win point serves as low hanging fruit that all teams can achieve, but that doesn’t directly effect the outcome of the match teams achieve it in. And for high level teams it also serves as a sort of consistency test to reward those who can still complete this task as well as do a goal rush consistently during qualifications.
That is pretty bad. Doesn’t happen in Michigan. But that’s not my point … I just don’t consider one line (go forward) an auton. I love encouraging them … but the game is already too heavy on the “remote control” side of things than the “robotics” side of things. I’m not sure setting the bar at one line of code pushes them enough.