View Full Version : 2014: the year of the underdog
Consider these classes of robots:
A) A box on wheels that does nothing but herd balls on the ground. Easily achievable by every team.
B) Robot A, with a "basket" on top which can catch a ball that lands in the basket, and then release that ball onto the floor to be herded. Not much more difficult than A.
C) Robot A, with the added ability of picking a ball from the floor and shooting it into the high goal. By virtue of this ability, robot C can also shoot the ball over the truss. Quite possibly much more difficult than robots A and B.
D) Robot C, which can also catch a ball. Likely extremely difficult.
The extremes:
An alliance of three A robots scores 37 in auto, and 31 points per cycle in teleop.
An alliance of three D robots scores 75 in auto and 60 points per cycle in teleop.
The middle:
An alliance of 2xA, C robots scores 57 in auto and 40 points per cycle in teleop.
An alliance of A, B, C robots scores 57 in auto and 60 points per cycle in teleop.
My take: 2014 is the year of the underdog.
In recent games, an alliance of three great scorers was virtually unstoppable, because multiple game pieces allowed simultaneous scoring. With one ball in play after auto, this advantage is significantly reduced. The only way an alliance of three D robots beats the A, B, C alliance is if their auto shot accuracy is PERFECT. If one D robot of the three misses its shot, the auto score for that alliance becomes 55, and all the extra effort that went into those three D robots is nullified.
It seems that scoring an autonomous goal by pushing the ball into the low goal should have quite high probability of success, and high goal auto shots will be significantly less reliable.
I could easily see three A robots, beating the pants off three D robots. The A robots could use "assembly-line" passing to run up the assist scores, and high probability low goal scoring to reduce cycle time. To keep pace, the D robots need high accuracy shooting into the high goal. If not, chasing rebounds and re-shooting will kill their cycle times.
The top level teams will try to build D robots, with varying levels of success. Depending on a team's resources, they should focus on a C level robot with extremely accurate shooting, or the more easily achievable B level robot. B robots with reliable low goal auto modes will be the unsung heros of regional competition. They are easily achievable, and allow an alliance with just one accurate shooter to be extremely competitive.
This year everything hinges on the accuracy of your high goal shooters. If they are nearly perfect, they win. Anything less, and the tenacious A and B robots will beat them.
If you are a team of limited resources, build yourself a drivetrain as quickly as possible, so your software team can get your low goal auto mode working perfectly. Then give it a basket to be a B robot. Make that easily swappable for an accurate shooter, and only make the swap if that shooter is darn near PERFECT.
This year will be the year of "A good robot built QUICKLY beats a great robot built slowly."
One final thought on the accuracy of high goal shooters. There are NO safe shooting zones. Last year you were safe to shoot when touching the pyramid. The year before, you were safe when shooting from the key. This year, good luck getting a clean shot off. Think you can maintain a high scoring to shot ratio when anyone can ram you while you're shooting?
Justin Shelley
04-01-2014, 23:45
I strongly agree that this game is built to even the playing field for all teams. Unless they add some sort of endgame mid season...
RRLedford
05-01-2014, 00:26
The opening post of this thread is an extremely astute analysis of this years game considerations, and the advice for the build approach is spot on, IMO.
Our team has decided to build the KOP frame & drivetrain ASAP.
In parallel we are planning to build a precision shooter based on a surgical tubing driven ram with an infinitely variable retraction winder.
We want to determine that we can reach a 90+% success rate on the high goal autonomous shots as soon as possible, and then proceed with detecting of hot goal, ball pickup, and other functions.
-Dick Ledford
Jacob Bendicksen
05-01-2014, 00:27
I wouldn't discount the mid-season endgame just yet, though it would probably be sooner rather than later. Maybe that will include a safe zone -- if you're touching the truss you're safe or something along those lines.
I entirely agree. I think this year, it will be very interesting to see how teams with limited resources keep up. I think this game will have everything to do with how fast you can get the ball down the field, and I think teams who throw it all over the place will lose track of the ball very easily (especially considering its relative lightness and bounciness(and varying levels of bounciness due to inflation)) and I think recovery from that is going to take a long time. Therefore, your 'A' and 'B' robots who theoretically have very controlled strategies could indeed do very well. However, I think its harder than we think to bulldoze around a ball due to its lightness and bounciness.
I also can't get myself to ever think of defense when I think of this game being played. That is a criticism of myself; I'm not saying that defense seems unlikely. This game has so many complex strategies, its hard to wrap my head around it.
Team3266Spencer
05-01-2014, 01:00
I entirely agree. I think this year, it will be very interesting to see how teams with limited resources keep up. I think this game will have everything to do with how fast you can get the ball down the field, and I think teams who throw it all over the place will lose track of the ball very easily (especially considering its relative lightness and bounciness(and varying levels of bounciness due to inflation)) and I think recovery from that is going to take a long time. Therefore, your 'A' and 'B' robots who theoretically have very controlled strategies could indeed do very well. However, I think its harder than we think to bulldoze around a ball due to its lightness and bounciness.
I also can't get myself to ever think of defense when I think of this game being played. That is a criticism of myself; I'm not saying that defense seems unlikely. This game has so many complex strategies, its hard to wrap my head around it.
I think not being capable of good defense could lead to some missed opportunities. Consider that since there will be only one ball in "the cycle of assists" once you've completed your third of the exchange you'll likely have some dead time to fill before the next ball comes, especially when working with a slower alliance. That dead time seems like a great defensive opportunity to me.
I don't disagree with you, but you should be careful. Teams with a D robot are usually high profile and high budget teams. If they have the resources, couldn't they make a universal robot. And if they notice a strategy begins to fail, they could just change and go back to beating the A, B, and C robots.
AndyBare
05-01-2014, 01:53
I don't disagree with you, but you should be careful. Teams with a D robot are usually high profile and high budget teams. If they have the resources, couldn't they make a universal robot. And if they notice a strategy begins to fail, they could just change and go back to beating the A, B, and C robots.
I definately agree with this. It is extremely difficult to add during competition, but way easy to strip a bot. If they plan on being 120 pounds with a D bot, by the time those teams get to their A or B bot, they'll be much less, possibly even giving them an edge over other A's and B's. I am not saying it would be wise to strip a bot in competition I'm just saying it could end up helping in the end, if this proves true. But by the time World's roles around, I think the winning alliance will have all this under wraps. Many D's will do extremely well.
So true, Todd. You wrote so well what I have been thinking all night. Could be an interesting to see what approaches teams take this year.
One thing I have learned from watching FIRST for many years is that the top teams are not just the top teams because they build the best robots. They're the top teams because year after year, most of them understand the game and plan strategies far better than the average FRC team. Considering the depth of misunderstanding of the rules and potential strategies that have been seen on this forum today, I don't see that changing, and I don't think the theoretical ability of less-sophisticated robots to keep up will, by itself, help "underdogs" all that much.
Todd put together a great summary of the possibilities this year.
I think we'll end up with more teams playing a ground game this year. There are too many chances to lose control of the ball when it is in the air:
- If you're trying to catch after going over the truss, a small bump to the receiver will strip them of the 10 points for catching, and the ball is loose on the field.
- If you are trying to shoot at the goal and get hit/moved, you miss the goal and the ball is loose on the field.
I suspect passing on the floor will be used with robots close to each other. Again, if you lose control of your ball, you're sunk.
-Eric
Upon further reflection, and a good night's sleep...
This year, the game IS defense.
Defense against shooters isn't blocking their shots, like in 2013 and 2012. Blocking shots is only necessary if the shooters have safe zones from which to park, aim, and shoot. In 2014, defense against shooters is hitting them while they are trying to shoot. The 2014 game is football where the quarterback has no offensive line to protect him, and pass interference against receivers is permitted.
From a strategic point of view, consider how much harder it is to construct an accurate shooter than a robot that just pushes the balls around. Much of that extra effort is negated when it's legal to ram a shooter as they are shooting. Any scoring action (high goal scoring, the over-the-truss caught pass) that requires precise shooting becomes easy to defend against. The uncaught over-the-truss toss slightly less so, because it requires less precision. For teams with limited resources, I would suggest that those same resources and effort be spent to build two simple pusher bots very quickly, and then those bots be used to practice rapid exchanges and scoring quickly into the low goals. (With occasional detours to ram shooters).
If an alliance has a higher score, effort spent denying the other alliance from scoring is just as valuable as scoring yourself. There is only one ball at a time after auto. If you are ahead, and prevent that ball from being scored, you win. No further scoring on your part is necessary.
And, in the blessed event that one of those fancy shooters gets a ball stuck inside (how often did that happen with basketballs in 2012 and disks in 2013?), you have it made.
In recent games, the multitude of scoring pieces made defense tough. The math was simple. Two good offensive bots beat one good defensive bot because the defensive bot could only hinder one scorer at once. This year, even an alliance with three good offensive bots can be stopped by a single good defender, because after the initial 3 auto balls, there is only one game piece in play at a time.
I don't think the theoretical ability of less-sophisticated robots to keep up will, by itself, help "underdogs" all that much.
Agreed. The key to an underdog team doing well this year is recognizing the high value in finishing an A class robot in 4 weeks and practicing for two weeks rather than finishing a C class robot in 6 weeks and getting no practice. The extra abilities of the C class robot are easily negated by middling defense, and have cost them two weeks of practice time.
I disagree about the 'year of defense' thought Todd. A robot that is pure defense is denying their own alliance 20 points from the 3rd assist. The time to score the goal (either goal) is when the 'defensive' robot goes to get the assist for their alliance.
I think it's really more a year of smart strategists and smart drivers. Paying attention to the opponents' real-time strategy will be key. Paying attention to which opponents have an assist on a given cycle is a must. Even paying attention to which side a ball is about to inbounded from is a must.
Edit - come to think of it, Aerial Assist is more like Ultimate Frisbee that Ultimate Ascent was. There will be some cycles where the opposing team will say 'wtf just happened?!'.
themccannman
05-01-2014, 12:54
This is the first game in a long time where the bottom goal is a totally viable option. You only lose 15% of your score on a max point cycle going for the low goal rather than high. Which means if you can decrease your cycle time by 15% by going for the low goal you have effectively made up the points lost form not hitting the high goal. In a 30 second cycle, if you take 4.5 fewer seconds to score in the low goal than the high goal, you have the same scoring efficiency.
I think that this closes the gap quite a bit. Although as stated, the tradoff for defense is that your alliance won't be able to get the third assist (20 points). Unless you can escape the robot you are defending, gain possession, then pass it off and return before the opposing alliance can pass to the defended, defense might just be removing two robots from the game.
Although because of the difficulty of shooting, defense on high shooters can be easily accomplished with a good drivetrain and a 4-5' tall robot.
omalleyj
06-01-2014, 07:14
<snip> The time to score the goal (either goal) is when the 'defensive' robot goes to get the assist for their alliance.
I think it's really more a year of smart strategists and smart drivers. <snip>
I disagree with your first point, because of your second point. Good strategy would be to always have the robot not involved in assists doing defending. Which means either the two most capable robots assisting each other and the third always defending, or any combination where the defender never breaks off until a replacement is almost there.
The only way to counter is two robots that can pass and receive faster than can be defended. I think this is unlikely unless its very top tier against a fairly slow defender.
Of course, all that said, you know some teams will find ways to score. They always do :)
Our team finished working through our analysis of the game yesterday. When I came back to CD this morning I was struck by the irony of the title of the "not so secret endgame" thread. It's ironic because the end game this year is not so secret. This year, the end game is cycling. This becomes more apparent when you do cycle time analysis, adding up the time required to perform the actions in a fully scoring cycle. The overwhelming time killer is chasing uncontrolled balls around the field.
This year's game has been fiendishly designed to tempt teams to shoot for the high goals during autonomous. But at the same time, the field is designed to make successful high goal shots in auto fiendishly difficult. The robots must start in the white zone. The white line is a long way from the goal. If you've tried it, you know that a successful goal from the starting position is difficult even for a person. Most robots will need to drive forward and get closer to the goal to make that shot, stopping in just the right spot. This isn't easy. I don't expect many will be able to do so. Let's say, one in three. And if there is only one robot which can do it, putting a goalie in front of them reduces their chances further. The net result is that high goal attempts in autonomous will result in many balls that rebound, often all the way to the far side of the field.
Let's say one of the three high goal attempts in auto goes in. That leaves two balls that must be retrieved and brought back for shooting. But, where do you shoot from? There are no easy references, like a pyramid, or a key. Plus, defenders are allowed to hit you while you are shooting. So let's say 1 of the two remaining auto balls makes it. That means you get to play "go fetch" AGAIN. Alliances which are enticed into the trap of only going for high goals will be lucky to have time to complete even one "cycle", let alone multiples. They will spend all their time chasing rebounds from missed high goal attempts.
Contrast this with an alliance of low goal scorers in auto. They might come out of the first 10 seconds with a lower score, but even the balls which aren't scored are still under the robots control, and can be quickly disposed of. Then, these three robots are ready to begin cycling IMMEDIATELY. And it's the triple assist scoring cycles that really run the score up.
omalleyj
06-01-2014, 09:27
<snip> The overwhelming time killer is chasing uncontrolled balls around the field.<snip>
The logical inferences from this is never losing control of balls; either by never transferring (low assist scores), or direct robot-to-robot hand offs that are difficult to defend without being penalized. But that means having other robots that are designed suitably in sufficient quantities. If one (or more) of the Ri3D efforts do this then maybe it could happen, otherwise...
(BTW, thank you for these comprehensive analyses and your very understandable write-ups)
cmrnpizzo14
06-01-2014, 09:34
This is the first game in a long time where the bottom goal is a totally viable option. You only lose 15% of your score on a max point cycle going for the low goal rather than high. Which means if you can decrease your cycle time by 15% by going for the low goal you have effectively made up the points lost form not hitting the high goal. In a 30 second cycle, if you take 4.5 fewer seconds to score in the low goal than the high goal, you have the same scoring efficiency.
I would actually disagree that the low goal is a viable option. The low goal is small. To actually score, you need to get right up next to the goal. Chances of shooting the ball in from any sort of distance is incredibly low. A good defensive robot could easily block one of the goals off from your alliance and then probably delay a score in the other goal. It will take much more effort to actually get to the low goals.
The high goals on the other hand are so wide that if you have a shooter that is decent you should be able to get to a spot to shoot within a short matter of seconds if you have anything comparable to a kitbot drive. Additionally, if you pick your shooting position correctly, a goalie should be a non-factor in blocking the shots.
I'm sure that there will be at least one team that proves me wrong but for the vast majority of teams I feel that the high goal is a much easier scoring option when facing defense in eliminations. The easiest comparison would be fender shooters vs. key shooters in 2012. Fender shooters were effective in qualification matches but they were neutralized in eliminations when defense stepped up.
The logical inferences from this is never losing control of balls; either by never transferring (low assist scores), or direct robot-to-robot hand offs that are difficult to defend without being penalized.
Yep. So our top two priorities are
1) A drive train that moves in auto.
2) A reversible floor pickup that picks up balls from the floor, or another robot (delivered at ~floor height) and spits them back out either into the low goal or to another robot with a floor pickup.
(Note that these are only the top two priorities. The list continues from there. )
With just these two systems, and good driving, you are a very valuable member of an alliance.
If I had the choice of picking this robot with a drive team who has practiced retrieving loose balls for two weeks, or one which had a high goal shooter but no practice time, I'd go for the first one.
Lil' Lavery
06-01-2014, 11:15
I would actually disagree that the low goal is a viable option. The low goal is small. To actually score, you need to get right up next to the goal. Chances of shooting the ball in from any sort of distance is incredibly low. A good defensive robot could easily block one of the goals off from your alliance and then probably delay a score in the other goal. It will take much more effort to actually get to the low goals.
The high goals on the other hand are so wide that if you have a shooter that is decent you should be able to get to a spot to shoot within a short matter of seconds if you have anything comparable to a kitbot drive. Additionally, if you pick your shooting position correctly, a goalie should be a non-factor in blocking the shots.
I'm sure that there will be at least one team that proves me wrong but for the vast majority of teams I feel that the high goal is a much easier scoring option when facing defense in eliminations. The easiest comparison would be fender shooters vs. key shooters in 2012. Fender shooters were effective in qualification matches but they were neutralized in eliminations when defense stepped up.
There is a lot of truth to this, especially given that protrusions into the low goals seem to be allowed (http://www.chiefdelphi.com/forums/showthread.php?threadid=124089) (GDC, please change this in Q&A/Team Updates!!!). However, the counter point is pretty simple, you only need one robot on an alliance to shoot into the high goal. Could there be benefits to having multiple? Absolutely. But for many of the "underdog" teams, focusing on being an incredible ball handler is going to be a better focus than shooting, imo.
What's REALLY promising about the low goal for underdogs is that the lip on the bottom of the low goal is 7" - which is the lowest height allowed for the top of bumpers. Thus getting it into the low goal is literally as easy as 'just run into the thing' - so long as a team can acquire the ball and then get to a goal.
If the GDC doesn't change the 'protrusions into the goal' issue, then I suspect there will be a LOT of broken protrusions...
Rypsnort
06-01-2014, 15:53
Todd put together a great summary of the possibilities this year.
I think we'll end up with more teams playing a ground game this year. There are too many chances to lose control of the ball when it is in the air:
- If you're trying to catch after going over the truss, a small bump to the receiver will strip them of the 10 points for catching, and the ball is loose on the field.
- If you are trying to shoot at the goal and get hit/moved, you miss the goal and the ball is loose on the field.
I suspect passing on the floor will be used with robots close to each other. Again, if you lose control of your ball, you're sunk.
-Eric
In regards to the small bump while receiving a ball over the truss stripping the team of a chance to get the points:
If the team is able to just herd the ball in a desired direction they will receive points for a catch. (I remember reading this in the manual once, but I am having trouble finding that part of the manual again.)
In regards to the small bump while receiving a ball over the truss stripping the team of a chance to get the points:
If the team is able to just herd the ball in a desired direction they will receive points for a catch. (I remember reading this in the manual once, but I am having trouble finding that part of the manual again.)
However, that herd has to happen before the ball touches the ground, which seems like it would be pretty hard to do.
However, that herd has to happen before the ball touches the ground, which seems like it would be pretty hard to do.
One could argue that any large change in kinetic energy and direction while the robot is moving could be considered carrying or herding. So a 'high-quality pillow' on an inclined plane which nearly stops the ball and lets it fall to the floor could be considered a catch if the robot moves at all while in contact with the ball.
While I agree Aerial Assist seems to be built to prevent elites from leaving everyone else in the dust, I think it will still happen, and I think it will happen in a way that ultimately is pretty discouraging to the weaker teams.
In recent posts, I've listed why I think an elite playing by themselves, with their two alliance partners clearing a road for them, will be able to score in the 150 territory.
In the predict week 1 scores thread, I outlined why I believe the average week 1 score will be about 73.5.
This means that unless paired with other significantly above average teams, the elites will likely tend to want their alliance to allow them to be the sole offensive machine, because the weaker teams will slow them down so much it reduces their total effective scoring ability, EVEN with the big bonuses for assists. 6 1-assist runs with a truss toss (20pts) and 4 2-assist runs with a truss toss (30pts) have the same total score.
themccannman
06-01-2014, 16:53
I would actually disagree that the low goal is a viable option. The low goal is small. To actually score, you need to get right up next to the goal. Chances of shooting the ball in from any sort of distance is incredibly low. A good defensive robot could easily block one of the goals off from your alliance and then probably delay a score in the other goal. It will take much more effort to actually get to the low goals.
The high goals on the other hand are so wide that if you have a shooter that is decent you should be able to get to a spot to shoot within a short matter of seconds if you have anything comparable to a kitbot drive. Additionally, if you pick your shooting position correctly, a goalie should be a non-factor in blocking the shots.
I'm sure that there will be at least one team that proves me wrong but for the vast majority of teams I feel that the high goal is a much easier scoring option when facing defense in eliminations. The easiest comparison would be fender shooters vs. key shooters in 2012. Fender shooters were effective in qualification matches but they were neutralized in eliminations when defense stepped up.
I'm not saying you should go for the low goal because you're right it's much easier to defend the low goal than the high goal. What I'm trying to say is that if the high goal is blocked it's easy for you to just drive into the corner and quickly dump the ball into the low goal. It's viable to go for the low goal if the high goal is being defended because you lose very few points. Only 15% of your points on a perfect cycle compared to 67% the previous two years.
... 6 1-assist runs with a truss toss (20pts) and 4 2-assist runs with a truss toss (30pts) have the same total score.
For a single bot, I don't think it will be so efficient to run down the ball after it flies over the TRUSS.
Ian Curtis
06-01-2014, 18:56
What's REALLY promising about the low goal for underdogs is that the lip on the bottom of the low goal is 7" - which is the lowest height allowed for the top of bumpers. Thus getting it into the low goal is literally as easy as 'just run into the thing' - so long as a team can acquire the ball and then get to a goal.
Given how few goals (average 1-1.5 per robot per match) were scored in Breakaway with a lot more balls, and how few hurdles were made in 2008 (at my regional the average hurdler could get 1 per match) I think "literally as easy as" are very dangerous words.
Canon reeves
06-01-2014, 22:06
if you think about it, each zone is worth 20 points, a class B robot could focus on a great feeder, and catcher and catch often, and do good in defense and protecting the shooter from defense they are just as important and don't need a launcher at all. But without good driving theres not much you can do. Very good year for all the rookie teams including 2 year teams.
Canon reeves
06-01-2014, 22:21
I would take the B robot and just add a very reliable ground feeder that could pass off effectivly. We are trying to design a feeder on a long base robot because a longer base is more stable, better for defense and can take more hits without losing ballance. I think there will be alot of sheering of feeders if its not designed right.
vBulletin® v3.6.4, Copyright ©2000-2017, Jelsoft Enterprises Ltd.