Chief Delphi

Chief Delphi (http://www.chiefdelphi.com/forums/index.php)
-   General Forum (http://www.chiefdelphi.com/forums/forumdisplay.php?f=16)
-   -   WEEK 1 (http://www.chiefdelphi.com/forums/showthread.php?t=104131)

XaulZan11 04-03-2012 11:33

Re: WEEK 1
 
How successful was camera tracking week 1? I know in the past there has been a lot of varability and trouble due to inconsistent lighting.

Andrew Lawrence 04-03-2012 11:33

Re: WEEK 1
 
Quote:

Originally Posted by davidthefat (Post 1138485)
Wait, does this mean that our robot with 14 inch wheels, a terrible kicker and a bridge lowering mechanism actually has a chance this year? I balance solo in 4 seconds, two bots in about 15 to 30, depending on the other driver. For pete's sake, we have a KICKER... Let's just say I'll focus on defense.

David, a robot with the abilities you described has a certain niche imperative to winning this game: A feeder. If you could hold one to three balls, and then kick them back to your alliance's side, you effectively starve the opponents of balls to score with and feed your alliance. I know I'd want a feeder bot on my alliance, especially one that does the bridge well.

I'm just curious; Are you wide or long (robot)?

JewishDan18 04-03-2012 12:03

Re: WEEK 1
 
Quote:

Originally Posted by XaulZan11 (Post 1138619)
How successful was camera tracking week 1? I know in the past there has been a lot of varability and trouble due to inconsistent lighting.

It worked great for Team 20. It rarely failed, but we suffered from other problems.

davidthefat 04-03-2012 12:59

Re: WEEK 1
 
Quote:

Originally Posted by SuperNerd256 (Post 1138620)
David, a robot with the abilities you described has a certain niche imperative to winning this game: A feeder. If you could hold one to three balls, and then kick them back to your alliance's side, you effectively starve the opponents of balls to score with and feed your alliance. I know I'd want a feeder bot on my alliance, especially one that does the bridge well.

I'm just curious; Are you wide or long (robot)?

It was designed to be a feeder, but I am highly skeptical of its ability to kick the balls effectively enough. It is a long bot. I have been very against this design, but I guess I see a silver lining. The main issue is that we will end up running over the balls, so we would have to back up to kick it. I rather not waste my time lining that up and trying to kick. It has no ability to "dribble" balls, or even hold onto it.

Talks Too Much 04-03-2012 13:09

Re: WEEK 1
 
Our tracking software didn't have any problems at HH, which took place in a gymnasium.

ColdRail 04-03-2012 13:48

Re: WEEK 1
 
Quote:

Originally Posted by pfreivald (Post 1138612)
:ahh: We didn't add anywhere close to 200 pounds... I think we added 120 -- but we used the battery test as indicated on the field tour videos.

Here's hoping!

SciBorg Dave and I mentor together on team 4061. I helped with the bridge calcs and tipper design....Actually we added 140 lbs to our bridge which weighed 104 lbs unweighted. That got us in the range of the calculated bridge weight using statics calculations for a total bridge weight of 244 lbs based on parameters from the battery test video. Once our bridge was weighted we never got the battery test to work exactly like the video but got very close. We felt this was likely due to differences in our hinge mechanism and minor variations in our wood bridge construction compared to the competition bridge design. We designed for a 2X torque margin (mini-van door motor and gearing) to make up for any calculational uncertainties as well as wanting to ensure that we did not blow any fuses during a match due to the tipping motor pulling excessive current.

In our own testing, our tipping system works very well with power to spare (imagine a robot doing push-ups or a low-rider car with hydraulic suspension hopping up and down!). The real test will be our first ever competition in Cheney Washington April 5-7.

pfreivald 04-03-2012 13:52

Re: WEEK 1
 
Quote:

Originally Posted by ColdRail (Post 1138695)
In our own testing, our tipping system works very well with power to spare (imagine a robot doing push-ups or a low-rider car with hydraulic suspension hopping up and down!). The real test will be our first ever competition in Cheney Washington April 5-7.

We tweaked our bridge so that it was harder to tip than the battery test would have made it, and our manipulator worked just fine. I'd claim I'm not worried, but I'm always, always worried!

davidthefat 04-03-2012 15:40

Re: WEEK 1
 
Well, it turns out that the top seed at San Diego is 4161. The do not shoot from what I hear from a little birdie at San Diego, all they did was coopertition.

littlejimmy1058 04-03-2012 15:44

Re: WEEK 1
 
Week 1 is never perfect!!

George Nishimura 04-03-2012 17:35

Re: WEEK 1
 
As far as I can tell, it's all about teamwork. Timing, possession and consistency.

Wayne TenBrink 04-03-2012 22:47

Re: WEEK 1
 
I learned that when you go into Week 1 at 119.8 lb, you can't add new stuff without taking something off. Hmmm. Software upgrades work. 1's and 0's are light.

A few of the things that would be useful on the robot:
- A good way to see under the far side of the bridge.
- A good way to clear balls from under bridges.
- A good way to tip the bridge to your side without any of the balls going under the bridge.
- A good way to tip the bridge for somebody else without blocking their path or having to lead them over the bridge.
- A better way to keep balls from getting rolled under the chassis.
- Shooter wheels that aren't affected so much by ball variation.

Joe Schornak 05-03-2012 01:07

Re: WEEK 1
 
Something that surprised me at San Diego was how different alliance selection ended up being in comparison to previous years, mostly because the coopertition bonus allows teams that cannot shoot baskets consistently to seed very high. In every previous regional that I have attended, the uberbots who could hang all the rings or score all the balls invariably seeded in the top eight, and alliance selection consisted of the 1st seed picking the 2nd seed, the 3rd seed becoming the 2nd seed and picking the new 3rd seed, and so on. At San Diego, no alliance's 2nd pick was from the top eight, since most of the really strong scorers had been too specialized towards shooting to do coopertition in many matches. This meant that the elimination alliances were more balanced than in previous years, and I did not get the impression that any one alliance was significantly stronger than the others.

Based on these observations and from looking at the team ranking charts, I have concluded that it may be more important to get coopertition points than to win a match if one wishes to be highly seeded, which strikes me as somewhat counterintuitive to my competitive sensibilities but probably reasonable. Has anyone calculated any sort of "sweet spot" for number of matches won without coopertition vs. matches lost with coopertition? Would it even be possible to calculate this?

On the whole, I really like Rebound Rumble. The qualification matches can be rather boring if everyone just stays on their own side, but the endgame is very tense and exciting, especially in the eliminations when an alliance goes for a three-bot balance and only makes it in the last few seconds. Near-realtime scoring is a very good thing, since I don't have to wait until the end of a match to see who actually won after penalties are calculated. I look forward to seeing how everyone adapts and adjusts in reaction to what we have learned this weekend.

Randomness 05-03-2012 01:18

Re: WEEK 1
 
Quote:

Originally Posted by Joe Schornak (Post 1139234)
Based on these observations and from looking at the team ranking charts, I have concluded that it may be more important to get coopertition points than to win a match if one wishes to be highly seeded, which strikes me as somewhat counterintuitive to my competitive sensibilities but probably reasonable. Has anyone calculated any sort of "sweet spot" for number of matches won without coopertition vs. matches lost with coopertition? Would it even be possible to calculate this?

It might be possible if you had a function that represented your scoring over time and the probability of balancing the bridge over time. Your expected win % and coopertition % for a given length of balancing could then be calculated.

Of course, the formula would vary from regional-to-regional (teams score different amounts at each regional) and match-to-match (it is probably best to spend more time balancing against weaker opponents.) I would use scouting, experience, and common sense instead of a formula.

davidthefat 05-03-2012 01:28

Re: WEEK 1
 
Quote:

Originally Posted by Joe Schornak (Post 1139234)
Something that surprised me at San Diego was how different alliance selection ended up being in comparison to previous years, mostly because the coopertition bonus allows teams that cannot shoot baskets consistently to seed very high. In every previous regional that I have attended, the uberbots who could hang all the rings or score all the balls invariably seeded in the top eight, and alliance selection consisted of the 1st seed picking the 2nd seed, the 3rd seed becoming the 2nd seed and picking the new 3rd seed, and so on. At San Diego, no alliance's 2nd pick was from the top eight, since most of the really strong scorers had been too specialized towards shooting to do coopertition in many matches. This meant that the elimination alliances were more balanced than in previous years, and I did not get the impression that any one alliance was significantly stronger than the others.

Based on these observations and from looking at the team ranking charts, I have concluded that it may be more important to get coopertition points than to win a match if one wishes to be highly seeded, which strikes me as somewhat counterintuitive to my competitive sensibilities but probably reasonable. Has anyone calculated any sort of "sweet spot" for number of matches won without coopertition vs. matches lost with coopertition? Would it even be possible to calculate this?

On the whole, I really like Rebound Rumble. The qualification matches can be rather boring if everyone just stays on their own side, but the endgame is very tense and exciting, especially in the eliminations when an alliance goes for a three-bot balance and only makes it in the last few seconds. Near-realtime scoring is a very good thing, since I don't have to wait until the end of a match to see who actually won after penalties are calculated. I look forward to seeing how everyone adapts and adjusts in reaction to what we have learned this weekend.

I just thought that it was just the nature of San Diego, we seeded 4rth at SD last year, we never hung a single tube, just did the minibot.

Well, we had to take off the ball manipulator on our robot due to weight problem... I am fearful of the mediocre shooters converting into feeder bots and playing defense.

Siri 05-03-2012 05:39

Re: WEEK 1
 
1 Attachment(s)
What I learned:
- As usual, Week 3 will be nothing like Week 1. I think this is exacerbated this year though, as originally teams had been very reluctant about defense given the number of safe zones. Even with two absolutely fantastic scorers (341 and 1218), shut-down defense let our alliance win by a wide margin in basically every elim match.
Stats:
- For all elimination matches combined, our alliance's average score difference was 49.2, all positive. This turns into 57.8 if you discount the QF in which 341 broke at the beginning--we won this by only 6 points. (They were up and running great next match without a timeout.) It drops to 44.3 if you ignore the (unsuccessful and I think unintentional) G25 red card our opposing alliance got in QF1-2 (their pre-card score was 29).
- For fender defense, the average score difference was 55 (47.8 without the red card, 71.3 without 341's breakdown, 61.7 without both). For the 2 QFs we played against an alliance of 2 dunkers, the average difference was 54.5 (adjusted 103, 40, 74 respectively). Actual scores were 41-47 (QF1-1) and 103-0 (QF1-2, pre-card 103 to 29).
- We also played fender defense in the semis against an alliance whose main offensive bot was a close scorer. This average score difference was 55.5, with nothing really out of the ordinary for which to adjust.
- We played what I like to call Inbounder Chicken in the Finals, blocking 2016's Inbounder throws and thereby hoarding balls on our scoring side and forcing 2016 (their long-range shooter) back to feed. (Also creates conditions for Key fouling, or Key-Ally mirror fouling which almost happened.) Based on the Finalist Alliances pre-finals teleop scores (average 15) as well as our alliance's same (average 30.7), blocking could have had a negative impact on the Finalists of approximately -6 and a positive impact on us of approximately +8.8, for a total differential difference of 15.2. Of course, there are also other factors, and granted we won by much more than 15 each match.
- I don't think Co-Op necessarily skews seeding too much, and this will likely decrease even more as teams get better at balancing. 341 and 1218 ranked 1st and 2nd whether sorted officially, by Teleop Points, by Teleop + Hybrid, by Teleop + Bridge, or by Teleop + Bridge + Hybrid. (Of course, they're also the 1st and 3rd OPRs in the country...) Overall it seemed that most strong teams realized the importance of Co-Op, and were ready to use it when beneficial (basically always).
Stats:
- In fact, at Horsham, sorting by TP+HP+BP versus sorting officially only produced a significant difference in 1 of the top 8 seeds, namely the 8th. 2234 would have been 30th, but had a CP of 8, matched only by 341.
- Overall, the rank difference averaged 7.7. 14 teams had a higher official ranking, 20 had lower, and 4 didn't move (seeds 1, 2, 3, and 29).
- There were two teams in the top 50% with 0 Coopertition Points: seed 5 and seed 12.
- Anecdontally, I actually found this to have a smaller impact than the minibot of last year. We ranked 2 of 56 (8-0-1) and won Philly that year when we would have ranked 18th (5-3-1) and probably not made it past quarterfinals without the minibot. Our estimated robot contribution was like 5. Of course, this is highly situational.

Quote:

Originally Posted by Littleboy (Post 1138098)
Exaclty how are they [fenderbots] being defending? Preventing? Pushing?

In elims, we finished autonomous and raced (pre-agreed route) over the Barrier, down the Ally, and along the fender. We stuck to it (front and non-Ally side) basically the entire game. While shutting down two dunkers wasn't exactly easy, it was entirely doable and they even got in each other's way at the beginning. They scored 5 points total in Teleop the first match, and 6 the second match. Scoring from the side was equally blockable--no advantage there that I can see. You can't just sit there to block, but I didn't let my driver lose bumper contact with the fender until endgame.

Quote:

Originally Posted by Deetman (Post 1138450)
From my viewpoint observing in the stands, the "quality" of the balls in play generally seemed to exhibit general wear and tear akin to what teams noticed during build season...

2018 (HH winning alliance, 2nd seed) had a unique system on their robot that measured the compression of the balls and adjusted their shooter accordingly.

Agreed. It seems the competition balls are actually more consistent in manufacturing quality than the ones we received in the Kit or FIRST Choice. However, they are this manufacturing style was also the minority of what we were given in the aforementioned. As far as wear and tear, the MAR field at Hatboro-Horsham was only given 36 balls total, but they still managed to remove basically any ball missing more than say 1/2in^3 of foam from one place. I was pretty impressed, actually. Mostly just scuff lines. We were given some of the decommissioned balls (the winning alliance can't cut nets at MAR, so they did this instead--thanks), and they do get very beat.

Yes, 1218 used the "charminator" (like terminator) to measure ball squishiness. As their OPR and Blue Banner can attest, they're a fantastic shooter because of it. 341 took a different tactic, putting huge (I mean huge) backspin on the balls so they fall basically straight down once they hit the backboard no matter what.

Quote:

Originally Posted by SuperNerd256 (Post 1138620)
David, a robot with the abilities you described has a certain niche imperative to winning this game: A feeder. If you could hold one to three balls, and then kick them back to your alliance's side, you effectively starve the opponents of balls to score with and feed your alliance. I know I'd want a feeder bot on my alliance, especially one that does the bridge well.

I have to say, watch out for this. Having just won an event spending all of Finals blocking an Inbounder and all of Semis blocking a short lower scorer, I'd warn that unless you're a high kicker this is pretty easy to defend. It'll be especially true in elims, but if you're good I'd expect defense to pick up in quals at later weeks as well.

Sheet 1 of the attached worksheet has the elimination match score differentials for the Hatboro-Horsham's winning alliance and the pre-finals teleop estimates for the Impact of Inbounder Chicken. Sheet 2 has the seeding differences with and without Coopertition.


All times are GMT -5. The time now is 16:11.

Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
Copyright © Chief Delphi