2012 Lessons Learned:The Negative

[vhcook]

I deeply dislike the new Q&A system. The number of clicks to get each answer in the new system when trying to keep up with all of the rulings was excessive. The new system's design seems to have an flawed assumption that we only care about the answers to questions we asked.

I really wish the Q&A system had an easy way to pull a digest of all answers (and their questions) since X date or last visit, since that would exactly match the reasons I go there.

I also missed having the referees summarize fouls at the end of the match. I like the instant scoring, but sometimes I was looking at the other end and didn't see the foul or referee hand signal happen.

[rocknthehawk]

Originally Posted by JuliaGreen

I would like to have written feedback from the Chairman's Award judges for every team that presents at a regional or CMP. There is a judging rubric out there - I'd like to see something that helps the students understand their strengths and weaknesses.

Julia

Is this not standard? I know we picked up our Chairman's feedback from pit admin after awards were over. (At GSR). It is hugely helpful, pointing out our strengths, and the one thing we didn't touch on. We won EI and will push harder for RCA next year

[nitneylion452]

Originally Posted by omalleyj

Ooops, Forgot my Kinect Rant:

Why was a device made to track humans put in areas where humans (other than the human player) could easily walk through? At every event (we went to 4) we had instances of people including photographers and field personnel walk through or into the Kinect area during hybrid. There needs to be something more than lines on the floor, especially when so few teams utilize the Kinect.

How about "Drivers behind the lines; confused bystanders out of the Kinect areas; 3, 2, 1, GO."

At the MAR Championship, we were blocking the Kinect area from the time we got the green light to the time hybrid ended. Our field supervisors made sure that nobody and I mean NOBODY (I had to keep some of the esteemed VIPs from walking through) through until hybrid was over. We made it a point that teams who used Kinect weren't interfered with during its operation.


To comment on the thread: I like the new radio (mainly because of the AP option for team use) but something with a faster boot time needs to be used. Also, the communication issues I witnessed at the MAR Championship were disgraceful. Teams sometimes spent 7 or 8 minutes with the FTA to get their robot connected, only to have it lose comms a few seconds in to the match (1676 comes to mind here). It was a shame.

Also, an issue I was personally affected by was the method to dislodge balls from the baskets. During the MAR Championship elimination tournament, alliance 1 (341, 25, 1640) clogged the basket in hybrid with no fewer than 6 balls. I was behind the basket attempting to nudge them lose, to no avail, so we had to restart due to a field malfunction. As I came out on to the field, I fake cried and said, "They just wouldn't stop coming! They just wouldn't stop!" I was told by 25's coach that they shot on something like 5,6,7,8, and 9 second marks, so I just had to be ready. I jokingly replied with "So I get .5 seconds to react? Sweet!" We both laughed and got back to the match.

Anyway, something to prevent the jamming of game pieces needs to be implemented any time something like this year's game is done. A pole to poke the balls wasn't nearly effective enough.

[Nemo]

On better webcasts: I totally agree.

To add to that, I think it would be useful to have some guidelines to give to all of the camera people on where to point the camera when. If you're trying to watch a particular robot, as most people probably are, it is very frustrating when the feed focuses in on the one robot that isn't working for 20 seconds at a time. Or if it focuses on a robot that is about to shoot, and then pans to another angle before we get to see the shots. Etc. etc.

Echoing what was said above, it would be great if FIRST would take charge of this and do it right: high resolution video, 2+ views including a full field view, archived and easily accessible according to event and match number, reliable live stream, etc.

That would have obvious benefits in getting more people exposed to FIRST. The grainy feeds where you can't even recognize team numbers, or even really see what a robot looks like, do a poor job of showing how cool FRC is. Those feeds are only really useful for people who already know how the game works and what the particular robots they're watching for look like.

Also... video feeds of sufficiently high quality would allow stat nerds to compile actual stats for every event. That would be sweet!

[Grim Tuesday]

Originally Posted by IndySam

The GP award made no sense at all. I bet there was nobody at most regionals that could even explain how it was determined.

Quoted for truth. I don't think the GP Award could have concrete criteria, or else they would have to make concrete definitions for GP, which is somewhat against the concept. At the same time, it should be about general GP of the team, and not about a single event (though it can be). Also, deciding on it before elims is silly. The greatest case of GP in FIRST tend to happen during eliminations. For example, at Champs this year, I would have given it to all the teams on Einstein for being gracious and professional about the whole ordeal.

The moment that made me realize what GP was about (I'll tell this story any time I get a chance) was in 2010 Philly, where an alliance of 341 and 365 called a time out for our 7th seeded alliance so we could fix one of our robots. We had already used our coupon, and our alliance was the biggest challenger to them. They called a timeout, giving our alliance enough time to make a C-RIO changeout and have a 3v3 match. We ended up losing 7-8 due to a penalty, but that gesture will always stay with me and is a testament to what its like to play against two HoF teams on one alliance.

On the subject of competition this season, I would have to say my biggest complaint is the manual. This years manual was vague to say the least. Glossary. There needs to be one. If the GDC insists that the game is played exactly the way they want it to, we would never see any of the really cool an unique robot designs. For that to happen, teams must have the freedom to know if their designs are legal or not. If the excuse for not giving specific definitions is that they will in Q&A (as was stated at FRC Live), they need to actually give the definitions asked for in Q&A. A second thing that struck me about the FRC Live is that they said that they were simplifying the manual "because some teams don't read it". We shouldn't reward them for not reading the manual, and cater to them at the expense of unique designs.

No internet/power in the stands was a big inconvenience for our team. Would be nice for FIRST to contract with whatever venue they have for internet during the competition.

Suggestion for FIRST: Send a recording device along with each field for archiving events. With their goal of bringing back alumni, it feels like they are forgetting old games relying on teams to record them. I can barely find any media of games before ~2004 or so, and FIRST has a much longer heritage than that. FIRST Official webcasts would be nice too, if their goal is more media coverage, this wouldn't be a bad place to start.

Overall, an excellent season.

[Racer26]

A few things:

Someone mentioned earlier in the thread the matchmaking algorithm. It patently doesn't suck. It's a very complex multidimensional matrix problem, and it does a very good job at producing a 'fair' schedule. It makes an assumption that all teams are equal in power, and thus, a given schedule will be 'harder' or 'easier' for some teams. It has a large number of parameters which can be set, including minimum match separation and so on. There's a well written paper explaining exactly how it works. Someone can provide a link I'm sure.

My biggest wish is for the robots to stop using consumer-grade electronics (the D-Link radio) in an industrial environment. It wasn't designed for the kinds of shock-loads and so on it receives in the hostile environment of an FRC bot, so it should come as no surprise that things like the power connector don't hold up terribly well to the abuse we put them through.

Einstein made it evident that we need to a) solve the FMS/robot comms issues, and b) have a much stronger plan in place for what to do when Einstein has systemic issues.

The people complaining about the CMP division size: get used to it. It will be the new norm unless they go to 8 divisions, which comes with the complication of where to put that many fields. We have reached a point now where the number of regionals qualifying teams to CMP completely fills the ~350team capacity CMP had pre-2012, and we're only adding MORE qualifying events with each passing year.

Better webcasts. Absolutely. A few this year, GKC, and GTREast, come to mind as having been pretty good. Perhaps better casting and archiving will come from the newfound Google partnership? After all, Google owns Youtube.

[Jon Stratis]

My negatives:
1. The bumper issues this year were horrible. Through all the weeks of regionals/districts, about half the teams at each event had bumper issues that needed to be fixed. That number is way to high to blame it all on the teams.
2. There was very little defense played at the regional level this year - for the most part, teams stayed on their half of the court and did their own thing. There were exceptions, but it made the game a little boring.
3. The game was a little more challenging than in past years, which I feel hurt the rookies. In the past, a kitbot could go in and do some good for a team, even if it's just driving to the other side and getting in the oppositions way. Not so with this year, as traversal of the field was non-trivial (despite how easy some teams made it appear!).
4. The Kinect has a great potential "coolness" factor that wasn't realized with this game. You didn't have long enough or accurate enough control to make it really useful during autonomous, and only letting one team on each alliance also limited its usefulness. This was something I really wanted to use this year, until I figured out we could do everything we wanted with a normal autonomous mode.

Some fixes for the above negatives:
1. FIRST has already stated they're getting a group together to go over the bumper rules for next year, and that group includes mentors and inspectors, the two groups of people that mater the most for this issue!
2. I'm not sure there's anything FIRST can do about this, other than to make sure we have a more balanced game next year!
3. I really feel that there needs to be something simple a team can do to materially benefit their alliance. I've been here for 6 years now, and tell tell you exactly what kitbots could do in every game up until now: Rack 'n Roll, they could play defense with little or no risk of penalties. Overdrive, they could do laps to score points. Lunacy, a good driver could avoid getting scored on. Breakaway, they could play defense fairly well just by getting in front of the goal. Logomotion, they could play defense really well just by pushing tubes to the sides and getting in the way. The game next year needs to include something for a simple kitbot to do!
4. The game needs to involve some way for the kinect to very clearly be superior. For example, in overdrive the balls were randomly positioned, and the hybrid remotes could, in theory, tell your robot where the ball actually was. The kinect gives us more control, so maybe take this same concept and make it 3 dimensional? What about allowing control of the robot through the kinect during teleop? There could be a portion of the field that isn't viewable from the driver station, but is from the kinect station - the human player can "take over" with the kinect in order to do something for some serious bonus points for their alliance.

[Alan Anderson]

VIMS needs a complete makeover from end to end. That doesn't mean just grafting additional functions onto the existing core.

[Craig Roys]

Originally Posted by Jared341

Having illuminated targets does not solve the problem that Don mentions. The retroreflective targets where just as easy to track (if not moreso) than self-illuminated targets...but in either case there is always the possibility that something off in the distance will be look the "same" to your vision system (in color if not in shape/size).

An opaque top backboard would have partially addressed this, and it's not like the audience seated behind the ends of the field was able to see anything anyhow.

I agree on the opaque backboard, but have to disagree about lit targets. You are neglecting the inverse square relationship between light intensity and distance. A reflective target forces the light to have to travel twice as far to reach the camera than a lit target (from robot to target and back to robot vs from target to robot), not to mention the fact that some of the light hitting the reflective tape will be scattered. Your target, being on the field, would almost surely be much closer to the camera than any other lighting in the venue making light from that target appear brighter than anything else. With a lit target, you could tune your camera to pick up only the brightest objects (your illuminated target) while ignoring most everything else. I would take a lit target in an arena filled with LED signs over retroreflective tape any day.

[Tetraman]

Originally Posted by Astrokid248

1. If someone posts a question, and you don't answer, and someone posts a more specific version of the question trying desperately to get some sort of a ruling, POST THE D@#N ANSWER. DO NOT say something like "we're not going to review your design" and then backtrack 6 weeks later. Along those lines, use common definitions of words and tell everyone what dictionary you're using, like Webster 7th Edition, so that all of us who aren't in the GDC can go look up the words and better determine how to design our robots until you answer our questions.

There are many, MANY reasons why the GDC doesn't answer certain questions on the Q&A, specifically about robot and mechanism designs. What the GDC didn't do was backtrack, they purely did what they should have done - avoided making a judgement on specific designs during build season, and then made a ruling during competition based on what was defined in the rule book. There were multiple choice words used in the Rule Book to explain what robots were unable to do, and all a dictionary would do is provide those same words back.


That said,

1. Allow for 'conversation' in the Q&A.
What I mean by this, is that a poster can respond to an answer they received within the same "thread" of questions. Rather than it being that a new question needs to be asked, but all questions relating to an answer can be grouped together and not separated by other questions.

2. Rules involving everything behind the player station walls need to be enforced and not ignored. Or, remove those kind of rules.
Example, the rules about the inbounders this year. Very rarely, at many events, these rules (don't step over lines, hold only 2, take out of corral asap) weren't enforced. Maybe in elimination matches the refs gave a little check that all was well, but for the most part, these rules didn't hold any ground.

3. Practice Rounds could be changed.
Instead of a list of matches that will be played by specific teams, have practice matches be first-come-first-served. This will only work best at regional and district events, not the championship; that said it makes more sense at regional/district events anyway. Have teams line up at que who wish to be qued up for a match and enter them in as they wish. Teams will still be required a time to connect to FMS on practice day, to ensure connectivity, but otherwise all teams that wish to forego their practice matches will then allow an extra team willing to join in a go.

4. Official FIRST Video Recording and Archive of all Competition matches, Alliance Selection. All Competitions webcasted by official FIRST staff/volunteers.
This, obviously, will take some time but it's so needed and pretty self explanatory. Done should be the days of teams having to volunteer coverage of events, and instead it be done officially at one website for all to see.

[coldfusion1279]

I think that the Q&A is the cause for a lot of gripes this year.

I would like to see it eliminated. They are digging themselves a deeper and deeper hole as FIRST grows. Seemed like the GDC couldn't keep up with aggregate thinking of the masses.

The rules should be clear enough to make the Q&A redundant.

[Steven Donow]

Originally Posted by coldfusion1279

I think that the Q&A is the cause for a lot of gripes this year.

I would like to see it eliminated. They are digging themselves a deeper and deeper hole as FIRST grows. Seemed like the GDC couldn't keep up with aggregate thinking of the masses.

The rules should be clear enough to make the Q&A redundant.

While that is true, inevitably, some form of Q&A is needed because people are ALWAYS going to have questions about rules unless the GDC releases a manual with blue boxes for every single rule detailing every possible game/design scenario possible...


My grips this year are redundant to what's been said in the thread:

Comm issues: This one doesn't need to be discussed further.

Lack of a glossary: This year's manual needed it. In addition to the Q&A responses, it would have been nice to have definitions of things such as "grab/grasp/grapple" and "bridge" so that all the controversies with the rules wouldn't have happened.

[Travis Hoffman]

Originally Posted by Nemo

I totally disagree. The number of times alliances cooperated with each other is way higher than the number of times multiple teams conspired to hose a good team. Fact: this year, lots and lots of teams cooperated with each other to gain a mutual benefit. At least in the case of our team, that resulted in many positive interactions with teams that we otherwise wouldn't have had a chance to work together with. I'd guess that the GDC's "great intentions" ran at least partly along these lines.

Also, look at all of the teams that won #1 seeds this year. They are consistently really good teams who deserved to be there, and the number of exceptions doesn't seem much different to me than it has been in past years. This indicates to me that it was either not that common or not that easy for teams to collude and hurt the best teams' rankings.

I agree with this pretty much completely.

Don't let the inactions of a few overshadow the positive interactions of the many.

[Grim Tuesday]

Originally Posted by Jon Stratis

My negatives:

3. I really feel that there needs to be something simple a team can do to materially benefit their alliance. I've been here for 6 years now, and tell tell you exactly what kitbots could do in every game up until now: Rack 'n Roll, they could play defense with little or no risk of penalties. Overdrive, they could do laps to score points. Lunacy, a good driver could avoid getting scored on. Breakaway, they could play defense fairly well just by getting in front of the goal. Logomotion, they could play defense really well just by pushing tubes to the sides and getting in the way. The game next year needs to include something for a simple kitbot to do!

If you build a small bot like 4334, (Archimedes champs) or 522 (NY Regional winner) you could be quite successful this year. A simple kitbot (on steroids) could balance on the bridges by following another team up, play "pushy pushy" defense or with a good driver, play "ball control defense". There was lots for a kitbot to do this year.

[Steven Donow]

Originally Posted by Grim Tuesday

If you build a small bot like 4334, (Archimedes champs) or 525 (NY Regional winner) you could be quite successful this year. A simple kitbot (on steroids) could balance on the bridges by following another team up, play "pushy pushy" defense or with a good driver, play "ball control defense". There was lots for a kitbot to do this year.

Agreed, this year there was really not much different a kitbot could do than in past years. In fact, because of balancing, it could be argued that a basic bot with no shooting or bridge mechanism would have more success than a basic bot last year that had no hanging or minibot deployment.

Also, it was 522 who won the NYC Regional/