I’ll definitely have to test shots bouncing out. I didn’t consider that (we just had a tube/hole for the high goal in the human-robot tests). And as for drive practice, it just scares me that we could be the best low goal bot and still score less than an average high bot due to a defender that knows what they’re doing.
After running a bunch of robot profiles (assumed cycle times, make %, etc.) through a spreadsheet, I agree going low goal is worth a look. An average lowbot’s make % might be 85% while an average highbot’s make % might be more like 40-60% (if that)… higher if close-in fixed shooting spot IMO… (Yes, elite teams will shoot high & at a high %.)
Cargo balls will be bouncing like crazy to the far reaches of the field, so average cycle time projections are tougher to make IMO, but on average cycles won’t be super short for most teams. Net is… it could be close to a wash scoring wise between low and high. “Shoot from anywhere” and “Shoot protected” are advantages of high, yes, if you can actually execute those things. If you went into build season saying, “we need a simpler robot this year”, then look at low.
Another consideration - in eliminations, the extra benefit of the Cargo Bonus disappears, so if you CAN build a reliable high shooter, it could have an advantage in the elims (or playoffs or whatever the bracket is called this year). However, I also agree with all of the points about less bouncing around, cycle times, and reliability of shots for the low goal. Hmmmm…
Personally, I don’t think going for the highest rung is worth the difficulty, but still considering the second-highest. Point values seem really well thought-out this year.
This is, IMO, the tricky bit. In 2020/21, for example, our home-built goals were a very good analogue for the true field. At present, I’m considerably less confident about this year’s home-built’s accurately modeling however much “bounce out” the real field will have. Our modeling indicates we should be able to do a very good job of getting balls into the goal from reasonable distances, but no idea how many will actually STAY in.
Not sure if anyone has mentioned this yet or not, but the upper hub opening is about 4.5x the width of the ball. It sounds like a lot, but I think it will be deceptively challenging to get a shooter that can consistently hit that shot from the launch pad. Idk though, it’s early on and I don’t think anyone has tested how these balls work yet.
But is it possible to reduce cycle times by half by scoring in the low goal? Assuming similar accuracy for both (though upper would of course be somewhat lower), that would be necessary to justify focusing on the lower goal.
Instead of thinking of low and high goal shooting as mutually exclusive, do low goal shooting when it is most beneficial. During autonomous there is no defense, and there is a huge reward for accuracy. If you have 3 robots shooting high in autonomous (possibly at the same time, so their balls will hit in midair) and ANY of the shots miss, you likely have zero chance of scoring 5 balls in auto.
If you think 100% accuracy in auto is easy, check the match video from the week 1 competitions in 2020 and see how often all three robots made all 3 of their preloaded balls in the high goal during auto.
As an alliance, would you rather come out of auto with 10 points on the board and 13 more goals for a guaranteed ranking point OR 12-16 points on the board with 16-17 more goals for that same ranking point? 13 goals divided among 3 robots is 2 ish cycles each (at 100% scoring). 16 goals is 3 ish cycles each (at 100% scoring).
My suggestion is first make a 100% reliable low goal shooter/dumper. Use that during auto and 2 cycles after. Then switch to your high goal shooter to run up your score. If your high goal shooter is >50% accurate, you have a net gain. If (as is more likely) your high goal shooting accuracy is less than 50%, stick with low goals. This strategy gives us a solid performance base from which to start our development efforts, and a fun development path with a concrete performance goal for our high goal shooter. Want to attempt high goals? Demonstrate in practice that the accuracy of the high goal shooter is greater than 50%, and you can use it in teleop.
You already get a 2s reduction in ball processing time. I don’t think it’s unreasonable that balls bouncing could take another 2-3s. This is pretty squarely a 50% cycle time reduction, even with pretty conservative estimates for bouncing. You could also make a case that the balls will travel far further post-exit when dropped from the high port, further increasing this advantage.
You don’t need to be cycling twice as fast as a high goal shooter to beat them as a low goal.
So lets assume a shooter has 75% accuracy in teleop (in my experience in shooting game this is a very accurate shooter), during the match they can run 4 two ball cycles (once again this sounds low but will most likely be more than early week winning teams can do) so they score 12 points.
I make a robot who does slam dunks no shooter, I just run up slam into the low goal and tip my hopper. That is 2 points every time since baring a freak accident I am 100% Accurate. This means to keep up with the high goal shooter I need to run 6 cycles to their 4 to keep up in points. I also have produced twice as many cargo for the RP. These extra cycles are made easier by the fact that as others have mentioned cargo comes out of the low goal quicker and more controlled then the high.
The big downside to the low goal, is that it’s so much fun to make a ball fly way up high in the air! so the team misses out on all that fun. And misses out on all the fun of trying to get the ball into the high goal. And by “fun”, I mean really difficult challenge.
Yeah … we timed the bounces at 8.5 s. But does processing/bounce time really matter if we don’t know where the ball will come out of? We might end up chasing balls around the field either way.
A bouncing ball is erratic and will be affected by more things and potentially more drastically by said things than a rolling ball.
If a bouncing ball hits one of the rings holding the field balls it can bounce 90 degrees in a different direction faster than you or your robot can react to it.
A rolling ball will either stop, continue straight or gradually turn in a way your robot can follow easily.
This is assuming the two balls in your next cycle are the same two balls you just shot. This will very rarely be the case. Even if all three of your robots have or just shot two balls, there are still 5 balls on the floor for you to pickup. Those balls will be further away from the goal than balls scored in the low goal, and there are only five, so low goal scoring does provide a cycle time advantage. But it is not as cut and dry as high goal = 2 extra seconds + bouncing time. Most of the time you will be moving from your shooting location to balls that have already come to rest, they were probably shot two cycles ago at least. This is not to say the low goal should be ignored, just that that cycle time estimate is flawed.
And if you are blocked either by contention (from 6 robots but only 4 low goal spots) OR by active defense?
If you think that most of the cargo you get will have been stationary for longer than 10 seconds most of the time, I have bad news for you.
Between missed shots, people messing up in-taking, balls being reintroduced and ball denial these things are going to be going all over the place. What I mean by ball denial is If I am playing on the “far” side of the field and either on my way to or near my cargo there is one of the opposite color you can bet I am hitting that thing to the other side of the field to make my opponents cycle times longer or their cargo harder to see.
When I said at rest I meant not bouncing, should have been more clear. I agree that the balls will be moving everywhere. Even balls that come out of the low goal will be moving, that is inevitable. A decent intake and driver will not care that the balls are rolling, bouncing is a different story. Also like I said low goal will increase your cycle times by keeping the balls close. I was only saying that calculating the cycle time difference by assuming you are picking up the same two balls you just shot is not accurate.
If all 6 robots are going for the low goal then in theory this is about coordinating with alliance partners. The other alliance can only contain 3 of them at a given time so 1 should always be available for each alliance, this is similar to 3 robots trying to work on the scale in 2018, you all need to discuss timing, priority and if someone would be better used pealing off someone from the other side from one of the spots.
Honestly I still think 6 robots full offense (this is an important distinction) fighting for 4 spots will still score more than 6 75% accurate high shooters because the low robots in theory actively don’t want to be there long. Meanwhile the 6 high shooters are less accurate (still) and can passively knock each others shots out, if you watched the field tour video when they show how much the cargo can bounce around up there I see a lot of shots that would have gone in this year being ricocheted out from one already in there.
Now from an active defense point of view if the argument is that the defender is circling the low goal area like a shark then once again at most their alliance is blocking 3 of the scoring areas and one is available the offense robot also has a slight advantage this year in that if they push someone into the hub then back out the robot who was pushed now has to go around the lower exits, and get back into position before the offense robot who is already passed said lower exits and is determining the direction of travel for both. Basically it is the opposite of rocket defense in 2019 since robots going for the rocket RP had a specific location they needed to go to and a small area to precisely score into, because of this defenders were able to park near a rocket and then peel robots who got past them off of it. While this year the defender does not know where the robot offense robot is heading (left, right, center) and once they get passed even if you attempt to peel the scoring is much quicker and more forgiving.
Now if you are arguing that the defense robot is chasing the low goal robot around and actively bothering them the whole time, then this is no different then the same argument for the high goal robot except now you have the added issue of not only pushing and bumping causing missed shots but also the height of a defender blocking camera vision or shots themselves.
Now for the high goal shooter one could argue this is mitigated by the safe zone, however unlike every other shooting game with safe zones in the last 16 years the safe zone is defined by you touching something vertical not horizontal. This is an important distinction because in other years you just needed some part of your robot, or just the bumpers inside of the volume of said area, this gives you a pretty big area to get away from defenders and get a clear shot. This year since it is vertical you must be actively touching it to be safe and it is only the one face of it, that makes denying access to the safe zone much easier than in previous years since you can just get to it first, or keep yourself positioned between them and the safe zone. It also makes it harder once you are actually there since at most you have 16 inches you can have the “body” of your robot away from but still touching the safe zone, and they can still just park 2 inches in front of you to block your visions or shots.
Cooper3175 has an excellent point. Everyone has equal chance to go after balls as they return. So it doesn’t matter how long your ball takes to return. This makes the 5 second vs 7 second difference irrelevant.
Also, the high shooter bots will likely be in a better position out on the perimeter to pursue returned balls and “rebounds”, since the low shooters must move up adjacent to the hub each time they score.
I agree mostly - where I disagree is that I think floor pickup will be essential to reduce cycle time and increase CARGO points. I wouldn’t be surprised if a lot of HP are busier during AUTO than TELEOP.