How many balls can you score in auto this year?

Technical
#3

I answered 1 but I guess we averaged slightly more than .5 due to some bounce outs and occasional other issues.

We did much better in autonomous than usual. I think it was due to a particularly strong programmer, and the overall improved state of the FRC software/hardware available. Command based was a struggle initially but paid off when we could string together and reuse commands for auto. Also, on the build side they put together a nice shooter, so it was possible/worth it to go for more autonomous points than usual. There was even talk of going for a 2 ball auto, which would have been unheard of in past years.

#4

Over 1/3 of responses saying 5 balls or more, but that doesn’t seem to line up with competitions. I feel like there is a lot of sampling bias here.

6 Likes
#5

This is Chief Delphi, so, yes.

12 Likes
#6

There is probably both sampling bias and a great difference between what works at the shop and on the field. I tried a 4 ball at my first event, but the most we got was 3 due to field differences and bounce-outs. Anyway, I’m really happy with how autonomous has gone for my team this year. Custom swerve auto movement has made us a lot faster.

2 Likes
#7

Our autonomous is better than previous years. We do 4 balls. In prior years:
2018(rookie): drove forward, dumped the cube on the switch, hoped it was the right side. In reality, our auto routine was to pull an NPE and crash our code.
2019: driver-controlled, scored our preloaded hatch panel 50% of the time
2019 honorable mention: this legendary play https://www.reddit.com/r/FRC/comments/l59luh/the_overenthusiastic_robot_or_check_your/
2020: 3 preload volley, move off the line. Never seen in comp.
2021 @H : we did do path following for the first time for the driving challenges.

2021 off-season: see 2020 but it did work super consistently.

This is our first time ever collecting game pieces in auto to then score.

5 Likes
#8

Same for 449. I feel like this is likely true for a lot of teams this year.

1 Like
#9

This is our most advanced year in auto yet, thanks in large part to learning command groups and the lower barrier to path following this year compared to previous years. We were the only team at our competition with a reliable auto that did more than 2 balls.

1 Like
#10

Our team has accomplished the best autonomous in our history.

At our last comp, after things were dialed in, our robot shot 5 balls once, and 4 balls many other times.

I think there were 2 or 3 occasions that our robot shot 4, and our HP shot/made the last one for a total of 5.

I think our swerve drive (and a lot of time getting path planning to work) really helped us accomplish this. The lack of bag day also really helped, as well as these game pieces being fairly forgiving.

2 Likes
#11

This is probably down to the fact that the staging of robots and cargo means a timed drive while intaking can give you an extra cargo.

2 Likes
#12

That’s been true in many years prior, though - and moreover, I’ve seen a bunch of intaking auto routines that interact directly with the human player.

I think we’re seeing a genuine improvement in tooling and team knowledge.

1 Like
#13

I feel like we are sooooo close to getting path following, I’m very excited.

3 Likes
#14

I am very happy to see the convergence of game design, COTS, and shared code leading to greater autonomous participation and contribution.

3 Likes
#15

One interesting effect: early in the season I wanted to set up drivebase simulation since we wouldn’t have a working drivebase for a few weeks. Turns out that involved adding basically all the path following infrastructure (besides actually defining trajectories). Last week I offered to add path following for a 4 ball auto. Ran sysid tests for about 15 minutes, went home and checked all the infrastructure, came back the next day with a working trajectory auto short of velocity P tuning.

2 Likes
#16

Better, I think? Before this I think our previous best auto was 2 hatch auto in 2019 with Hab 2 start.

This year we struggled with auto for our first comp, did better second comp, and should have 5/6 ball and Citrus style 2+2 auto ready for DCMP. There was definitely a learning curve but PathPlanner and WPILib have made things a LOT easier.

2 Likes
#17

It took me about 2 months to get PathPlanner up and running, but wow it works wonders. Programming complete 5 and 6 call was possible in maybe 3 hours? Now that I got the framework down I’ve been making complete routines and big changes at comp between matches.

1 Like
#18

I would say not quite as good due to how fussy getting balls to stay in the hub is this year (esp for teams w/o access to a real goal before real competition).
2019 - not really an auto (telop via camera)
2020 - five and six power cells autos
2021 - all the power cells (12? 13?) on our side
2022 - 2 and 4 cargos autos (although all 4 don’t necessarily stay in the hub yet)

#19

Well, it’s a bit like scouting. When you ask people what their robot can do, you’re going to get the best they ever did. For each aspect of their robot. Listen to the pit folks and most will tell you they can do a 5 ball, shoot on the move, and climb to the traversal bar.

The reality usually disappoints.

It takes time for people to learn realistic evaluation and for them to become introspective enough to admit what their actual performance is. Heck, even most experienced folks tend to overestimate the ‘average’ number of balls they get per match, because they discount those matches where they barely score anything.

We “can” score 5. We have only done so 1 time in competition, with 3 being much more likely. Our average is probably at 2 or lower due to dead matches.

#20

I would agree with this completely. We’ve written routines in previous years where game element pickup was “simple” (e.g. Infinite Recharge) but we didn’t trust them enough to fire them in real competitions.

This is the first year that we’ve actually incorporated it into competition code.

Prior years our autonomous has been pretty reliable and predictable, but simple. This year it is equally simple, equally reliable, but it is the first time we’re score more than just our pre-loaded game piece.

We have been using a nearly 100% reliable two ball, and as we go to DCMPS next week, we’re going to have a (hopefully) reliable three ball.

Our initial design goal on the software side was a 4-ball option, we’re not getting there before the end of the season but it’s going to serve as a great off-season goal and exercise.

2 Likes
#21

Bias, yes.

But, we have 3 different 2 ball autos as well as a 4 ball auto and a 5 ball auto. We have run the 4 ball auto in a couple matches, but most of our matches ran one of the 2 ball autos based on coordinating with our alliance partners in terms of who would do what in auto.

I answered ‘4 ball’ to the survey. I could have answered 5 ball even though we have not run that in comp at this point (only on the practice field). I could have answered 2 ball given that our most common auto was 2 balls. But the question was “How many balls can your robot score…” not “How many balls does your robot typically score…”. So looking at the typical competition performance of any given robot may not provide you with insight as to what that robot is actually capable of. I’m guessing a lot of teams will answer this based on what they have done in their practice space rather than what they have actually done in competition. The question is not that specific.

2 Likes
#22

Our auto capabilities are night and day compared to 2021. At an infinite recharge offseason event we were running autos based entirely off time, and didn’t even have encoders mounted on the drivetrain iirc. This season we have encoders on anything that might need them, well fit pathfollowing, and nothing is using time.

1 Like