Week 1 Robot Performance Poll

Given that we’re officially 2 weeks from the start of the competition season, I’m curious to know how everybody thinks their robots will perform.

Assume your robot performance at a week 1 event. Use estimated average for Teleop Cargo, and Maximums for Auto Cargo and Climber Level.

# of Cargo Scored in Auto
  • 0
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7

0 voters

# of Cargo Scored in Teleop (High + Low)
  • 0
  • 2
  • 4
  • 6
  • 8
  • 10
  • 12
  • 14
  • 16
  • 18
  • 20

0 voters

Level of Climb (1=Low, 4=Traversal)
  • 0
  • 1
  • 2
  • 3
  • 4

0 voters

1 Like

I’m surprised to see such high estimates. Our robot currently physically exists as a drivetrain and a bunch of pretty good, but un-integrated, prototypes; our CAD is 97% done and our 1st robot parts are maybe 25% machined. Our first competition is Week 4 and we planned out our season and milestones accordingly (although I’ll grant that we are a few days behind that schedule).

By Week 1 I expect us to have our robot built and have made decent progress on testing and trouble-shooting, but not have ironed out all the bugs yet. We’ll probably have an auton path that allows us to shoot the pre-loaded ball, our drivers will probably have had very little practice, and a lot of pre-programmed button controls won’t be very optimized yet. Very different from where we’d be if we’d designed our season around being ready for a Week 1 comp. I suppose having your first competition be Week 4 is somewhat unusual, but given that most teams aren’t competing Week 1, I’m surprised so many think their robot will be ready to do so well by then.

4 Likes

This post makes me feel so much better! We are in the same boat, maybe even a bit further behind. I have no doubts that we will have something ready for our first event, but it’s going to be down to the wire! :grinning:

Chief Delphi, the FRC discord, and any other online FRC communities will almost always be a non-representative, competitive-skewed sample.

Even with that said, I think the the cycle estimates are high. I will be very very impressed if teams are averaging ~5 cycles in week 1. Maybe the top seeds will be. I’ll be excited if lower ranked teams can do 2 cycles.

I think the climb average is pretty fair, with a balance between medium and high climbs. Medium climbs are pretty accessible this year with COTS solutions, and we’ve seen that high/traversal is way more achievable than where we thought we started (2013).

Number of balls in auto I think is also pretty skewed. At a week 1, I am betting on over half of teams to not have any auto past maybe leave tarmac. The top half of teams will have a 1-2 ball, and maybe the top 4-6 teams will have a reliable 3+ ball.

In general, I’m going to be betting against high success in week 1 events. We all overestimate the skill of an average FRC team, and that is only compounded by very few FRC events happening since November 2019, 38 months (or about 25 build season lengths) ago. If you have it, check your scouting data from your last normal regional. It’s not as rosy as you’d hope.

10 Likes

Ditto!

I think we can get 1 in auto (maybe not consistently) because we will have one pre-loaded. Teleop will be rough. Maybe I’m ambitious on the climb being done, but we’re not going for anything spectacular.

1 Like

So we made a conscious decision back in 2020 to still pretend the bag/crate still existed and we kept that for this year. So our robot needed to be “done” (it is never truly done until our last offseason) by bag day. Well we also opted into a Week 1 event so we accelerated our time table by a week and we actually are well on track, if not slightly ahead.

Now I know we are not by any means an “average team” just posting some data.

Looking at out internal data at our first event (Week 3) in 2019 we were averaging 1.2 hatches in auto, 5.8 pieces in teleop and had an average level climb of 1.8. For 2018 our first event (Week 2) we were scoring 1.4 cubes in auto, 5.3 in teleop, and climbed 100% of time, and brought a buddy along in 90% of matches, so I will make this number 1.9.

Ironically this kind of shows a level of consistency so that is what I went with to answer these questions (rounding up to the next whole number). 2 Cargo in Auto, 12 in teleop (6 cycles), and level 3 climb (we are ignoring the low bar like many teams so the level 3 is our “2nd” level).

The fact that at this moment the average auto is higher than us (2.51), while the teleop is much lower (8.5) leads me to believe maybe a lot of confident programs who are not confident in their drivers have answered lol

I would guess that most people voting are being very very optimistic, or voting like they are the only robot on the field. There are only 11 balls of each color on the field, and they are scattered after scoring. My guess is that there will be far more chasing down balls and scoring a single ball at a time than most guess. Defense bots will likely also be fairly powerful, if they can get the ball starvation technique down. Not to mention that many many teams are struggling just building a baseline robot this year, after having so much time in-between.

I predict that at least at 1 week 1 event, there will be an everybot that wins as alliance captain.

lol

12 Likes

We used to have a strategy lead who would walk around muttering “optimism is my enemy.”

Feels like some combination of optimism and sample bias. Our team lost 1.5 years of experience across the board. We just got back in our space ~45 days before kickoff.

I suspect many teams are dealing with similar losses, and that will affect average reliability.

3 Likes

Any of y’all have a heuristic for translating pit scouting? If a team tells me they can do 10 cargo in teleop, I’m probably dividing that in half.

Why ask? Let their performance speak for itself.

3 Likes

It’s called “match scouts say they only score 2, in a good match”.

(Probably won’t be quite that bad, but still…)

1 Like

I can guarantee we will score zero gamepieces, and will not climb, in week 1.

10 Likes

Anything you can verify in a match should be ignored in pitscouting imho, ask what they have in terms of functionality, and look at things like bumper quality, and wear

3 Likes

Ditto. At least officially.

Oh absolutely. I think my pitscouting sheet says “motors and wheels”.

The heuristics are nice in situations where you don’t have data on hand.

1 Like

my hack if you ask is look at the bumpers
badly built bumpers =0.25 * values
well built bumpers =0.5 * values
well built worn bumpers =0.75 * value

3 Likes

I agree. I’ve found a picture to be the most useful thing about pit scouting.

That said, It is an excellent opportunity for our students to go out, talk to other teams, and see what they did. I also like our students presenting our robot to other teams.

1 Like

“How are their bumpers” is another great heuristic.

I think I’m gonna make a thread for this!

The difference between “can do” which should generally be heard as “has done once” and “will do”, or “will average” is significant.

To be fair, it was not uncommon for WWII pilots to overestimate their successes by 8:1 or 10:1

2 Likes