PDA

View Full Version : [YMTC]: How many Co-Opertition Points?

Tristan Lall
03-02-2012, 02:17 PM
http://img542.imageshack.us/img542/815/48837432952attemptedbal.jpg (http://imageshack.us/photo/my-images/542/48837432952attemptedbal.jpg/)

Assume this is the condition 5 s after the end of the match.1 Everything is at rest.

Is the co-opertition bridge balanced?
Which robots are fully supported by the bridge?
How many co-opertition points should be awarded to each alliance?

You make the call!2

1 I don't know exactly when this photo was taken. It looked like this on the webcast in real time as well.
2 This one isn't hypothetical, so we're going to be second-guessing the officials. There's nothing wrong with that, as long as we leave insinuations about their competence out of it. (This was match 22 at Alamo (http://twitter.com/#!/frcfms/status/175644635866214400).)

BigJ
03-02-2012, 02:28 PM
1. No. Per The Arena 2.2.5 (emphasis mine):
A Bridge will count as Balanced if it is within 5° of horizontal and all Robots touching it are fully supported by it.

2. It appears to be 3743 and 488, can't see behind them.

3. 1 per [G51].

EricH
03-02-2012, 02:55 PM
I would say that BigJ is correct--I don't think there's any way that one goes to 2 CP/alliance. Not Balanced, 2959 is the reason because it isn't fully supported by the bridge.

Tristan Lall
03-02-2012, 09:23 PM
My turn: there are many problems with this situation.

Firstly, [G41] is broken:If a Robot from each Alliance is Balanced on the Coopertition Bridge when the final score for a Qualification Match is assessed per Rule [G37], each Alliance earns 2 Coopertition Points. If the Coopertition Bridge is not Balanced, but a Robot from each Alliance is fully supported by the Coopertition Bridge, each Alliance will earn 1 Coopertition Point.
Bridges get balanced. Robots do not. Let's agree to interpret that phrase as "If a Robot from each Alliance is on a Balanced Coopertition Bridge...".1

The definition of "balanced" is hidden away in section 2.2.5 of "The Arena" (not even in a numbered rule). It states:
A Bridge will count as Balanced if it is within 5° of horizontal and all Robots touching it are fully supported by it.

The bridge is clearly not balanced (2952 is touching/supported by it, but is also supported by the floor). So right there, 2 CP is off the table.

Now what about 488? Isn't that their bumper resting on 2952? 2952 is not fully supported by the bridge, thus neither is 488. And if even one iota of 3743's weight is supported by (the not-fully-supported) 488, then 3743 is not fully supported by the bridge either. So even 1 CP is not likely.

So why did the Twitter feed say 2 CP? (Shouldn't the FMS push a correction out if/when the match results are revised?)

And what about the standings pages? They're utterly wrong. For match 22, all teams should have received the same number of CP, because CP don't go away ever. (Even if you're DQ'd you keep the CP which count towards the award.) Instead, 2 blue robots and 1 red robot got the points, and the rest got 0 CP.

As for the amount itself, earlier today, those that got points had 2 CP; it was revised to 1 CP. I think it should probably be 0 CP.

In summary:

Obviously not balanced.
Probably none; small chance of 3743; minuscule chance of 3743 and 488.
Probably zero; minuscule chance of one.

And both of the score feeds are wrong for different reasons (and independently of which robots are balanced).

1 I won't even dwell on the libertarianesque capitalization scheme (http://volokh.com/2012/01/23/tim-thomas-libertarian/).

Andrew Lawrence
03-02-2012, 09:31 PM
What was 2982 doing? Was it an accident?

Tristan Lall
03-02-2012, 09:31 PM
What was 2982 doing? Was it an accident?
Trying to assist...and then couldn't get out of the way in time.

EricH
03-02-2012, 09:47 PM
Now what about 488? Isn't that their bumper resting on 2952? 2952 is not fully supported by the bridge, thus neither is 488. And if even one iota of 3743's weight is supported by (the not-fully-supported) 488, then 3743 is not fully supported by the bridge either. So even 1 CP is not likely.

I don't think that's quite correct...

3743: Completely, 100% supported by the bridge. Eligible for CP.
2952: 50-50 bridge-floor. Not eligible for CP; they are also on 3743's alliance.

488 is the tricky one. They are definitely on 3743 (completely supported by the bridge via 3743), and definitely on the bridge. From the angle of this picture, it's unclear whether they're even touching 2952; it almost looks like their bumper is going into a gap in 2952's bumper.

You're going to have trouble showing, just based on this picture, that 488 is not fully supported by the bridge. You've got to show first that 488 is contacting 2952, and then you've got to show that that contact involves 2952 supporting 488. I don't reasonably see that happening from this picture. Now, if someone else has a picture from another angle, or a clearer shot, that's different evidence.

Based on the picture, and based on 488 not apparently being supported by 2952, 1 CP.

Tristan Lall
03-02-2012, 10:56 PM
I think there are two separate, but related items for discussion.

3743: Completely, 100% supported by the bridge. Eligible for CP.
This one is kind of a fine distinction, but FIRST has been consistent in the Q&A that the whole weight of the robot must have a load path through the bridge to be supported (though not necessarily directly).

But the system is statically indeterminate, so the exact load path is unclear. If all of 3743's weight is borne by its own direct contact with the bridge, then it's supported. But given that it's in contact with 488's (flexible, elastic) bumpers, and there are numerous small protrusions that appear to be making contact, there's basically no plausible way for the load path not to pass through 488 at some point, and without at least some of the force at the contact point being due to 3743's weight. (Let's leave aside the question of whether 488 is supported by the bridge for a moment; we'll deal with that too.)

There's no escape hatch in the rule for this case. Given that it's not possible to formally determine the load path, the safe assumption to make is that contact implies support—because at a basic physical level, that's a substantially true statement. It also fits the rules just fine, because contact is irrelevant if both robots are clearly supported through the bridge.

Contrast that with the alternative—that load paths be assessed in every situation, and some judgment be made as to what degree of force to ignore. (1 N? 1 nN?) It's essentially unfathomable that referees are prepared to make that call consistently.

488 is the tricky one. They are definitely on 3743 (completely supported by the bridge via 3743), and definitely on the bridge. From the angle of this picture, it's unclear whether they're even touching 2952; it almost looks like their bumper is going into a gap in 2952's bumper.
I'm definitely unable to make the call with the same level of evidence as the referees on the field—so if indeed 488 and 2952 are not in contact, I've got no problem with that outcome.

If I'm correct on this second question, the first question is moot: 0 CP. If I'm not correct on this point, then there's a chance for 1 CP, if I'm also incorrect on the first point. (Of course, given the state of the rules, I don't think there's any way—other than an arbitrary interpretation not specifically addressed in the rules—that the first point is not valid. If that's the determination that the referee made, it's a stretch, but plausibly within their discretion.)

Of course, none of this negates the problems with the rule construction or the FMS feeds.

EricH
03-02-2012, 11:25 PM
.

But the system is statically indeterminate, so the exact load path is unclear. If all of 3743's weight is borne by its own direct contact with the bridge, then it's supported. But given that it's in contact with 488's (flexible, elastic) bumpers, and there are numerous small protrusions that appear to be making contact, there's basically no plausible way for the load path not to pass through 488 at some point, and without at least some of the force at the contact point being due to 3743's weight. (Let's leave aside the question of whether 488 is supported by the bridge for a moment; we'll deal with that too.)

I don't follow this line of reasoning. 488 is on top of 3743. If there is any reason 488 is not on top of 3743, then you need to get another picture before you post again.

If any load path of 3743 is passing through 488, it is going upwards. Therefore, 3743 is supporting 488. You cannot logically make the conclusion that because 3743 and 488 are in contact, 3743 is not fully supported by the bridge, without presupposing that 1) 488 is not fully supported by the bridge and 2) any contact between two robots on a bridge causes one of them to be supported by the other.

I shouldn't have to remind you that two robots being in contact does not necessarily mean that one is supporting the other. If that's the case, I would figure that many of the balances this weekend need to be negated due to robot bumpers touching. Therefore, one of your two presuppositions is out the window based on precedent and simple engineering. This does not negate the other, but it does make it very difficult to make the argument that 3743 is not fully supported by the bridge.

To do the test another way, and the classic FRC way: If 488 was to be removed from the bridge, would 3743 change position? (Considering a 6WD drop center's rock on shifting CG not to be a change of position here.) In the context of this picture, 3743 would not change position unless it rocked to the other side of the robot.

Some of those older definitions make a lot of sense right about now...

Tristan Lall
03-03-2012, 01:19 AM
I don't follow this line of reasoning. 488 is on top of 3743. If there is any reason 488 is not on top of 3743, then you need to get another picture before you post again.

If any load path of 3743 is passing through 488, it is going upwards. Therefore, 3743 is supporting 488. You cannot logically make the conclusion that because 3743 and 488 are in contact, 3743 is not fully supported by the bridge, without presupposing that 1) 488 is not fully supported by the bridge and 2) any contact between two robots on a bridge causes one of them to be supported by the other.

I shouldn't have to remind you that two robots being in contact does not necessarily mean that one is supporting the other. If that's the case, I would figure that many of the balances this weekend need to be negated due to robot bumpers touching. Therefore, one of your two presuppositions is out the window based on precedent and simple engineering. This does not negate the other, but it does make it very difficult to make the argument that 3743 is not fully supported by the bridge.
That's kind of what I mean by a fine point: it's not obvious, because we ordinarily don't pay much attention to small deflections and small forces. They just don't matter for ordinary strength and fit calculations, so we just ignore them.

But FIRST has not specified a scale, and at small (e.g. microscopic) scales, every real-world surface is uneven, is deflecting and exerts small amounts of weight. Additionally, things stick together at these scales, so small amounts of weight can be supported in both tension and compression. That's why, in the real world, contact implies support with unfathomably large certainty. I grant that's not perceptible—so let's put that aside for the moment. Maybe FIRST intended the support to be visible—although that's far from what they said.

At much larger scales, you can perceive many things which you wouldn't ordinarily care about. Here I'm thinking of things like deflections of bumpers. And before you object, consider the precedent: if any robot part, no matter how insubstantial, touches the key, you're considered in. As a practical matter, the referees limit themselves to what they can see, even if it's not a major piece of the robot. So why not here too?

Is there not one point along the length of bumper-to-bumper contact where 3743's bumper fabric visibly weighs down on a fold of 488's fabric? A pair of interlocking weaves in the fabric, or the edge of a bumper number being pushed back? Or more concretely, what about that red plastic panel on 3743? Isn't it pressing on the (full-width) bumper of 488? What about the rocking action of 3743's drivetrain? Or indeed given any component in contact with 488 at all, a slight tilt is all it takes for the weight of a component to contribute (slightly) to the normal force.

I'm contending that even if you draw the line at what's visible, the likelihood of some weight being transferred through 488 is very high (even if the magnitude of that weight is very small). And if you draw the line at the microscopic, it's essentially a certainty. I'm also contending that there's no reason in the rules or "technical common sense" to arbitrarily pick a threshold larger than the visible (e.g. specifying a quantity of weight that triggers the support rule).

Another way of looking at the ruling comes down to playing the odds. If you want to take the sure bet, you have to treat contact as implying support, unless otherwise demonstrated in a particular circumstance. If you want referees to have arguments about how much support is enough to trigger the rule, and whether that support is present, you can go with a different standard.

As for negating balances, FIRST ruled that when a robot is on top of another robot (in whole or part), and both robots are otherwise supported by the bridge, they're fully supported by the bridge. So I wouldn't be too concerned—this would not affect the outcome most of the time.

In any event, if 488 is not touching any robot but 3743, the question of support goes away—however I premised my call on the assumption that 2952 is in contact (and that such contact is obviously weight-bearing).

To do the test another way, and the classic FRC way: If 488 was to be removed from the bridge, would 3743 change position? (Considering a 6WD drop center's rock on shifting CG not to be a change of position here.) In the context of this picture, 3743 would not change position unless it rocked to the other side of the robot.
The real test is: If 488 was to be removed from the bridge, and substituted with an equivalent force distribution on the bridge and on 2952, would any part of 3743 change position? (You can't neglect the rocking, because it implies that 3743 was being supported.)

That's a familiar test, and decently easy to implement in practice, but it's not strictly what the rules call for today.

So to put it all in perspective, if we say that contact implies support, the only thing that changes is the situation where a robot is in contact with a robot that is not itself fully supported (and where robot-robot support is not already obvious).

The other takeaway point is that we don't have to perceive the amount of weight being transferred—only the support which inevitably leads to a quantum of weight being transferred.

EricH
03-03-2012, 01:59 AM
We've gone into the "zipties are a load-bearing surface" argument from 2005. If you have to lawyer the rules that much to show that a given arrangement is incompatible with the rules, you're probably going too far in. Common sense interpretation, anyone? (Oh. Wait. This is the FRC GDC we're talking about. Define grapple, please. Oh, and while you're at it--what would a reasonably astute observer think on seeing this picture? Obviously, two different things from two different observers.)

The refs called it balanced. That I disagree with--the bridge is not balanced because there's a third robot in contact with it (and it's not an opposing robot on an alliance bridge, which would make this entire discussion a moot point). However, lacking any evidence that one of the robots on the bridge is in fact in contact with, let alone supported by, the robot not on the bridge, that's 1 CP per alliance.

P.S. Tristan, I fail to see how the load, from 3743, is transferred to the bridge via 488 while 488 is on top of 3743. You're dealing with frictional forces, maybe--but if there isn't any motion, friction is a moot point, as whatever frictional force is available is more than enough to stop any impending acceleration.

qzrrbz
03-03-2012, 04:37 AM
Although we haven't seen an "aircraft carrier" type of robot yet, the interpretation of all robots fully supported by bridge extended to all robots must have all wheels on the bridge would make that technique useless.

I would prefer believing that all robot mass is ultimately supported by the bridge, which means you obviously can't have have your wheels on the ground, you can't be touching the ground, but doesn't rule out the you can be sitting completely or partially (as is this case) on top of somebody else technique.

Here, remove 2952 from the picture, and the answer is clearly "+2 coop" even though 488 is partially atop 3743. The restrictive idea of "not all of 488 is supported (directly) by the bridge" negating this is not "a reasonable observer's" interpretation. Add 2952 back in, and place 488's bumper on 2952's bumper as it seems to be, and you get some physics component of 488's mass directed to the floor through 2952's frame and bumpers == 0 CPs. That'd be my call.

03-03-2012, 08:29 AM
For what it is worth, 2952 was not touching us at all.

I was there. :)

Tristan Lall
03-03-2012, 05:15 PM
For what it is worth, 2952 was not touching us at all.

I was there. :)
I defer to your physical presence...I had only the photo and webcast to go off off. So, in that case, 1 CP (two fully-supported robots on an unbalanced bridge) is the right call—and the discussion Eric and I were having is not material to the outcome of the situation.

As for Eric's point, I think it's fundamentally a matter of interpreting the rules in a way that lends itself to consistency, rather than one that feels familiar. The rules and the Q&As propose no threshold for recognizing the support of a robot's weight, and I don't believe it is a good idea to pick an arbitrary threshold that doesn't at least have some physical or practical significance.

In terms of physical significance, I feel that contact implies at least an infinitesimal degree of support, in essentially all conceivable situations. (That's the idea that at a microscopic level, there's adhesion, irregularity and deformation—some of which is bound to be slightly weight-bearing.) So that's one standard we could use—and it's very easy.

Another option is for the referee to observe the conditions that could lead to weight being transferred (any member is deforming under contact and its own weight; any tilted member in contact implies there is a weight component in the normal force), even if the referee can't estimate the amount of weight involved or even if the amount of weight is minuscule. Static indeterminacy makes this somewhat more complicated, but you can still conceive of a probability distribution that implies that in all but the edge cases, there is at least a modicum of weight transfer through all members. This is a valid standard as well, because it's reasonably easy to enforce with minimal scrutiny, and has both practical and physical significance.

I think this rule is enforceable in a systematic way—indeed, the real problem with the rule is not in its construction, but in its interpretation.

Peck
03-16-2012, 12:16 AM
If u want to get really technical:

The gravitational attractions the robot causes and any magnetic forces provided by the robots near location might be causing an upword acceleration to the robot thereby holding weight.

Of course, The same argument could be used to say that the earths magnetic field and the gravitational attractions of the roof, and everything above it as well as the magnetic attractions caused by the lights and the wires...

Might I recommend not trying to get down microscopic in your arguments so this all doesn't come into play since the roof could very well be lifting as much weight as any robot.