Sack Attack Game Breakdown, Diagnosis, and Robot Design

Hey guys,

This is a paper in which I breakdown this year’s VEX game, identify several successful methods of playing, and theorize a robot that I believe would be very competitive.

It’s kind of long, but hopefully an interesting read.

Do you agree with my stance on Sack Attack? Did I get something wrong? Would the speculated robot be successful/unsuccessful at a high level?

I look forward to the hearing others opinions.
Regards, Bryan
New Compressed (zipped) (1.1 MB)

I’m glad to see that someone else agreed on the idea of sharing the motors between lift/drive!
I only skimmed through it, and I’ll need to go more in depth when I have more time, but you seemed to focus on hoarder bots a lot. From what I’ve seen hoarder bots, at least at our earlier competitions, have been very unsuccessful, it’s just way to easy to play defense on them no matter what robot you have.
What are your views on getting around defense?

Yep, once I narrowed my sights down onto hoarders I go into more detail with that design than any of the other “molds”. If I didn’t focus down the paper would be much longer.

There’s actually an entire section on the paper based around defense buried somewhere in there. Basically, I counter by putting a lot of effort into being able to maximize the robots ability to push. (~20lbs empty, and ~40lbs full) It also retains the ability to score very quickly like a normal robot which allows versitility in match if there is a lot of defense played. The math to show why this this more efficient than a smaller capacity design is also somewhere in there.

This is my kind of document! I have roughly the same views which I illustrated in a document of my own (unpublished), but it is not NEARLY as detailed. Good work, sir.

This is fantastic Bryan,

I just read it cover to cover and you did an excellent job. I’m not sure that this is the design that will take the World Title but it would be very competitive.

This is now on the list of documents I’ll show my team each year. It’s a great breakdown of the game and a robot to play it; also an excellent example of using CAD to refine a design without fully modeling the robot.

One concern would be with the design of the wedges, you didn’t draw them, but you have to be careful about potentially getting called for intentional tipping if you are pushing other robots with them.

Also with the power takeoff system you would completely remove power from your drive train. I imagine some type of break pad would be needed to stop opponents from getting in a well timed push and interrupting your dump. Pneumatics would add weight but would work well here, many teams did this last year.

The other issue I see that you mentioned but never addressed is teams defending the dump. Even with an 8 motor arm if a team sticks their arm over your robot from the side it would be very difficult for it to complete the dump, I’ve seen reverse dumpers defended this way. Currently, you don’t protect the other side of the trough so a clever team could catch most of your dump and be supporting the sacks at the end of the match, even if they aren’t able to hold them or move them to their own trough.

Thank you for sharing this with the community.

Shouldn’t you be able to theoretically just push their robot out of the way? Protecting your sacks is a good idea :slight_smile:

Very good document by the way, wall bots would be interesting to play…

A very good design document - well done. You do diverge partway through to elaborate on the design of a hoarder bot though, but it’s at least a good paper on those design choices. Hopefully it’ll inspire some teams to go about designing more like this.

There’s a discussion in some other thread about whether you should be having a low skirt to push sacks away. I won’t go into it, but my opinion is that it’ll get stuck, no matter how well built it is.

This is a thorough coverage of most of the different aspects of play (before you get into detailed design work), but it does miss out on the 2 vs 2 dynamic of gameplay. It’s not easy to think about large scale robot positioning and “field strategies”, but it’s a ridiculously important strategic feature (we have won matches against better alliances with positioning). I’m not sure how strategy influences mechanical design, but there is a big difference in gameplay when all 4 robots are on the same side of the field.

This is amazing :). I had very similar results after analyzing the game(of course i would disagree in some places). However, I noticed a few strange things though, such as sayings sacks are 4x4 inches.

Thanks, I’m glad you all took the time to read it.

Wow, that was just a really bad screw-up on my part. Obviously the sacks are 5x5".

I guess the paper doesn’t so much teach about the design process as it does provide an example of one way to do it.

I read that thread (I read most of the threads but rarely post.) There is a huge difference between having a low chassis and having wedges at a slight angle to the ground, being pulled down by rubberbands, and actually riding along the ground. While I completely agree that the sheer weight of many sacks on the wedges could stall a lighter robot, I have a hard time believing that a robot with this amount of torque would have a problem. If the sacks were somehow still able to get under the wedges then I agree that the next best option would be to simply rise the ride height.

My lack of coverage on 2v2 situations was semi-intended. In FRC (and VEX) we do LOTS of work to understand how the game dynamics will work out. However, we play with so many weak teams in qualifications that while designing we basically assume that we are trying to win a 2 on 1 match. If our partners can help us, great! If they can’t, that’s alright too. Once you have a robot that performs to this level you can stratigize to your heart’s content to beat other equally or better performing robots/alliances.

Wow, I’m flattered to be included on your list. Wedges ability to tip a robot is dependant on how long and how curved they are. Wedges which arc upwards as they get closer to the center of the robot are much more prone to tip an opponent than one which is linear all the way through its length.

I didn’t talk about it much (if at all?) but my thought was to stab a pneumatic cylinder either into both back wheel’s spokes or into the ground to prevent being pushed while the lift is being raised/lowered.

I completely agree that I mostly ignored defending the dump. Origionally in the paper I was going to show two iterations, one in which I eliminate the roller wheel in the front and add those motors to the drive/arm while maintaining the same descoring functionality (which is a pretty cool design). The second was going to be the passive trough cover that expanded as the lift rose and prevented robots from being able to put stuff in the trough to block/catch the sacks being dumped and made sure they all went into the trough. I also had part of this robot modeled in 3D. However, I decided that it was long enough as is and posted it allowing me to concentrate more on several designs I have going for FRC. As the robot stands in the end of the paper the best way to strategically dump would be to drive under the trough pushing the potential defender away and then dump. As to just sticking something over the lift… It already lifts with more force than most robots weight (meaning it would tip the robot over before it stopped lifting. If this is really a concern the lift could be further geared down. Its currently optimized for speed but I recall it being able to go up in roughly 2.5-3 seconds with 75ish lbs of force with the correct ratio.

Regards, Bryan

the issue with that is if you are 1v2ing
and the other alliance is strategically competent, it doesnt matter if you can outscore them 1:5 if one of them is a dedicated “blocker/humper” and the other robot can freely score

and before you say, “we will drive around them”, lets just say its extremely HARD to get around a robot thats trying to block you (assuming same drive ratio)

My thought on a regular robot trying to block an efficiency robot is that the efficiency and still scores because it is very difficult to block every sack the entire match and stop them from getting to the trough once. They will score significantly less so that it is hopefully easy for the teamate to win the match against a handicapped opponent.

Well, it matters a little bit. Theoretically, if we can outscore the opponent’s scoring robot 1:5 and the Defense robot can’t cut our scoring down to 1/5 of what it normally is then we still win. With that in mind the robot in the paper was designed to play with defense on it every match. I would agree that it is very difficult to go around a robot that wants to block you no matter what their drive ratio is. That said, lets pretend that instead of trying to score in auto mode our robot picks up 15 sacks. Now at the beginning of teleop the robot can push with ~28 lbs of force. If we really want be be somewhere we can push the defensive robot wherever we want on the field so long as they weigh less than 28 lbs. No matter what their drive ratio is their wheels will break traction with ground and/or trip their breakers before the heavier robot (us) proceeds to push them where ever we want. And forget about them pushing us anywhere. Wedges plus 28lbs heavy makes it imposible for them to really budge us at all. The defenders challenge is compounded by the fact that their are sacks everywhere on the field for us to pick up, and the more sacks we pick up the heavier we get and the harder to stop.

A second factor is that matches are never played 1v2. Even if our partner can’t score he can always play defense on the opponent’s robot which forces the opponent into the same situation we’re in but (probably) without the same pushing ability. If we can outscore them 1:5 without defense we should be able to outscore them if we both have defense.

Very impressive, Bryan. This is the type of analysis we love to see and hear about.

Thanks Karthik. I should probably take this time to apologize to everyone for any spelling or grammatical errors. Re-reading it I found a couple obvious ones that I somehow passed over.

Give my regards to 1114.
Thanks, Bryan

on the left side
at 1:40 you can see essentially (one of the best robots in the competition) being blocked off by a “rookie” robot/team (all gr 9 team) for the remainder of the match
as long as you have the “parallel wall” set up, they woudnt even get close to the troughs (2k could have dumped on the floor, but they won that anyways)
and turns out that WASABI was doing the same thing to the blue alliance on the right side, which made the game after 1:40 essentially stalemate
wasabi scored some in the process of blocking because they are that good (another “elite” robot in the region)

yes theoretically you can push
but remember your drive motors overheating may or may not be an issue from all the constant pushing
also, teams with pneumatic “brakes”, from last year at worlds, are essentially unpushable

and even if you could “push them where ever you want”, all they need to do is stay between you and the trough (as outlined in the above vid).
even if you do indeed have the pushing power, all they would have to do is lift their arm as they get pushed to the trough

indeed its possible for you to “sneak” a few points here and there, but what im trying to get across is that even a “semi-decent” robot can essentially stalemate an “elite” robot

which would be desirable to the alliance that isnt as strong

Interesting discussion,

If the wheel slips on the ground before the motors stall then we can push with reckless abandon. It’s then impossible to trip the drive motors even by pushing against a wall for an extended period of time.

Brakes are interesting. The idea behind brakes is to take power out of the pushing/being pushed equation which leaves only CoF and Normal Force. These are proportionate so a robot with a high CoF and a low weight may be able to exert a force equal to a robot that weighs more but with less grippy wheels. The limiting pushing factor for the opponent in our case is weight because we are using the highest grip wheels available. If they weigh less than the force we can exert (which is our weight) then we can push them. Because of this brakes that lock the wheels won’t help. Even wheels that can’t spin will still break traction (as is the case when pushing a robot from the side.) Brakes that deploy into the ground are more interesting. If they deploy something like a pneumatic cylinder into the foam they are now mechanically linking themselves to the field which means you’d have to tear up the foam to move them (because the force to move them has shot up to whatever it takes to rip the foam. It’s arguable that by getting a wedge under one side of them you might be able to tilt the robot enough to disengage the floor brake but that’s going by more of a robot by robot basis.

I’ve seen that video beore as well. The drivetrain that team emulated is clearly from 44’s 2010 robot. 44’s drive never had to deal with and thus wasn’t designed to handle defense. This team probably never practiced with any defense and as a result didn’t know how to handle it. It doesn’t help that they couldn’t physically push the other robot but there was at least twice when they could have escaped to the other trough.

You are absolutely correct that so long as said defensive robot stays between us and the troughs we can’t score. Thankfully, there is a multitude of ways to get around even the most aggressive defender. In FRC, we train our drivers to do “dodges” (coined from lacrosse.) One move we have we call “pushing through.” In this case the defensive robot is set up sideway like in the video. The offensive robot basically rams the defensive robot on its corner spinning it and allowing us to maintain forward momentum. Another good one is to set up parallel to the defensive robot and drive forward turning into them which results in both of you arcing toward the trough next to each other. If this is performed right, the end state would be you at the trough with the defender next to you. If the defender tries to head you off then they are no longer between you and the troughs. Alternatively, you could probably just push your defender diagonally into the trough then do a backwards arc to the other one to score in.

There are lots of options and I defiantly wouldn’t be so quick to say that a “semi-decent” robot can shut down an “elite” robot with a powerful drivetrain and well-trained drivers.

Regards, Bryan

interesting discussion indeed :slight_smile:
didnt talk strategy in a while

are you saying with the highest traction wheels, you can push a wall (wheels slip) all day long without tripping your breakers?

i find that hard to believe for a full “competition robot” (even with no sacks)

yes i was talking about pneumatic brakes at worlds
which would lift the robot with a bar of c-channel wrapped with anti-slip mat probably the highest Cof you can achieve

and yes, the wedge idea would be robot-robot based and im uncertain of the technical legality? (may want to ask that)

yes, i agree that this isnt the greatest example with the said robot being X holonomic, there are still many other videos where this strategy is demonstrated with “standard” tank drive vs another “standard” tank drive
in last years gateway, we specifically did a series of “blocking” tests and found that if the driver knew the strategy and knew what he had to do, it was highly, highly unlikely that the robot can sneak past
one exeption is if the trapped robot had a “special” mechanism thats specifically built for this situation
im expecting similar results, but there may be new variables involved with sacks and a “wall” of the entire field length (as opposed to the 1/2 field pinch point from gateway)

the driving techniques you have described, i have actually used some back in the round up days.
to a skilled driver, there are definitely tricks and maneuvers that can be preformed in attempt to escape from the “trap”
however, you must compare apples to apples and also assume that the defender robot’s driver is JUST as skilled, and although i dont have working drive bases available to me at the moment, i am sure that to every maneuver you have described, there exist a counter-maneuver to a driver thats equally skilled

after finals next week, i will see if i can run some more tests on defence with working robots

I’m not going to post confusing excel sheets full of numbers but basically pushing against a wall indefinitely is possible and is common among FRC robots with a low gear. It’s just a function of weight, CoF, power, and gearing. if the required slip torque for the wheel is less than the wheel’s stall torque then the wheel will always slip before it stalls. Because the wheels won’t ever have to turn with more force than it takes to slip the wheel stall current can’t be reached. If the required slip torque is far enough below the wheel stall torque then the motors will not ever pull enough continuous current to trip the breakers allowing a robot to indefinitely push against a wall without angry motors. The robot in the paper has a wheel stall torque of 3.02Nm. At 20lbs (empty robot) the robot’s required slip torque is 1.55Nm which is substantially less then its stall torque and allows us to push without ever really stressing the motors. As we collect sacks the robot becomes heavier and requires more force to slip the wheels. When weighing 39lbs (holding ~38 sacks) we reach the threshold where the wheels will no longer slip before the motors stall. At that point the required slip force (3.03Nm) becomes greater than the wheel stall torque (3.02Nm). Without reducing our pushing ability these numbers can be improved by putting the last two 393 motors into the drivetrain increasing the overall power available and/or further gearing down the wheels at the expense of speed.

To your point I’m not sure that this was possible in past VEX competitions as robots weren’t able to use as many 393 motors. Putting 80% of our (as of this year increased) available power into the drivetrain is the only reason that this robot can achieve such awesome pushing force while maintaining a very competitive speed.

As was clear in your video Holonomic drives are basically impossible to make traction limited in VEX because the motors within the drivetrain aren’t coupled. Each wheel’s stall torque is only that of one motor rather than four like our drivetrain. If you are interested in finding out how a robot needs to be geared to become traction limited with only 4, 6, and/or 10 motors and a variety of weights I would be happy to post those on here as well.

I think we’re starting to go in circles a little regarding defense. Countering counters to counters is getting a little silly. When push comes to shove, you push and shove. Gateway was an easy game to play defense in. In general it’s easier to get around someone the more room you have to maneuver. Cutting the maneuvering space in half basically makes defense twice as easy and is why robots could be successful last year corralling their opponents and keeping them there. If this holds up then defense should be twice as difficult to play this year – especially because the scoring objective runs the length of the field.

You also mentioned special mechanisms to get through a blocker. What are you referencing? Wedges would seem to perform this function since they make the opposing robot twice as easy to push as they would otherwise be.

If you have the motors, materials, and time it would be very cool to see how an eight 393 motor (normal speed) robot geared up 12:18 (12t on wheel) in high strength chain and with some 5-hole-wide-plate-steel-flip-down-wedges performs in practice. I think you might be pleasantly surprised.
Regards, Bryan