Your Team Sucks At Defense: Here's Why

This post is unofficially part of an unofficial series of posts. See the OG post by @Katie_UPS on Scouting and @tjf’s post on Programming.

Hi, my name is Jeremy and I like seeing robots hit each other. I’ve been 5012’s drive coach since 2014 with prior drive team experience on 399 before that. I’ve had the pleasure of coaching my drivers through playing defense and had the opportunity to witness some top-tier defensive play firsthand. You might know us from such smash hits as 2019 OCR SF1-2 and 2016 LA F-3. sorry 330 :heart:

I’m a huge fan of defense. I have a lot of opinions on it. Here are a few reasons why you might not be playing defense as effectively as you could be-

  • You didn’t read the rules
    • Any doubt in your knowledge of the game rules will hamper your ability to play the game. Know the rules and be confident that every move you make is within them.
    • Always watch the refs. Referees watching your robot won’t assign penalties without signalling first. If you see a referee begin to signal, take appropriate action to avoid those penalties.
    • Learn the referee hand signals! There’s a difference between a pinning count, a foul flag, and a tech foul flag!
    • Communicate with the head ref early and often. Ask clarification questions at the driver’s meeting if you aren’t sure of anything. If you ever get flagged, request clarification on why / what you can do to avoid it next time.
  • Your robot isn’t robust enough
    • Mechanical - your robot has to be mechanically robust enough to dish out hits and take impacts without risk of failure.
    • Electrical - your electrical system should provide adequate strain relief, shock mounting, and redundancies. The looming threat of a radio reboot or other failure is not something you want during a match.
    • Bumpers - :snake:
    • ANY doubt in your robot’s reliability will affect your driving, period.
  • Your robot isn’t built for defense
    • In the past, one characteristic of a good defender was always high traction, high pushing power, and max weight. The top level defensive metagame for a while in the 2000s and early 2010s centered on T-bones and pushing matches.
    • Nowadays, excellent defenders should be prioritizing quick acceleration over just raw pushing power. You’re looking for sprint distances as short as 1-3 robot lengths so that you’re accelerating to speed faster than your target.
    • Mecanum and Omni drives aren’t necessarily bad at defense on their wheels alone, but they might not necessarily be suited for all types of defense.
    • Make sure your robot can’t break any volume/size constraints while playing defense. The nuclear option is to completely remove any mechanisms that might violate any size constraints while defending, but it’s often an easy enough fix to prevent this in most robots.
  • Your driver isn’t comfortable with the robot.
    • Does your driver often start moving the wrong way only to change directions and waste time?
    • Do they overshoot turns?
    • Are they constantly at 0 or 100% throttle / no in between?
    • The solution is driver practice! Get them stick time year round, with a strong emphasis before/between events. Choose a control setup and lock it in. Make it so that the robot is just an extension of your driver’s movements. Your driver should not be spending mental bandwidth on how to manipulate the controls to get the desired output.
    • It’s important for the driver to carry the robot’s momentum smoothly from point A to point B. you waste valuable seconds decelerating and accelerating more than you need to.
    • Be honest with yourselves about the capabilities of your robot. Lying and inflating numbers helps no one.
  • “Defense is just bashing into other robots.”
    • There are lots of options for defense. Not all of them will work in every situation. Knowing when to apply each technique is the mark of a good defender
    • Robot-oriented/contact defense - useful for preventing a single robot from accomplishing its objectives. Effective in 2019 in matches where only one robot was capable of a solo rocket,
    • Goal-oriented/positional defense - for preventing any activity near a goal or specific area of the field. Useful in 2019 in situations where one rocket is almost full and the target robot doesn’t have enough time to finish the other rocket if blocked.
    • Counter-defense/enforcer. This style of defense is particularly potent if your partner(s) can guarantee completion of an objective(rocket completion in 2019, full court shooting in 2013, death cycles in 2014, etc). This usually involves a combination of contact and positional defense.
    • Game piece starvation - dependent on year. Team 16’s 2012 robot on Einstein was an excellent example of removing game pieces from your target’s side of the field to starve them of scoring opportunities
    • Game piece choking - the opposite of starvation. This involves getting game pieces where game pieces don’t belong. One early example of this was 987’s championship winning play in 2007 where they placed a game piece in a position to prevent their opponent’s endgame. Current rules and game design make this harder, but it’s still possible. 2016, 2017, and 2019 are examples where the game pieces were annoying to drive around and could almost be deadly if grouped up in the wrong place.
  • You don’t have a clear defensive goal in mind
    • One target - one robot or one goal. You’re going to be far more productive if you focus on one target or goal versus trying to bounce between multiple targets. By trying to split your focus between two different targets, you won’t be defending either as effectively as you could be defending one.
    • You don’t have to stay in motion! Sometimes effective defense means maintaining a strategic position and not constantly moving, especially if you’re playing goal-oriented defense.
    • If you’re playing robot-oriented defense, by all means do your best to stick to your target in a position that will hamper their scoring.The objective here is to force them to waste time/game pieces. A good rule of thumb is to try to stick to the business end of your target with a priority on always being between that end of your target and their current goal/game piece/etc
    • Be proactive, not reactive. Most defensive maneuvers are made well in advance of their target’s actions. Don’t chase, make your target react to you.
  • “Defense is for suckers, we’re too good to play defense”
    • 973 2011 - fastest minibot in the world, top tier scorer, played defense en route to their first championship win
    • 2014 - only one game piece in play per alliance means only one robot per alliance is doing any scoring. What do the other 4 robots on the field do?
  • You aren’t communicating with your alliance partners.
    • If your alliance agreed on a plan and you go off-script without communicating with your partners, that’s a big red flag. Some of my nightmare matches this season have included partners who didn’t communicate when they malfunctioned and went to go play defense or a partners who have strayed from a counter-defensive strategy without letting everyone know.
    • Communication goes both ways. If you’re broken, let your partners know. If your partners are struggling, ask if they need help.
    • My favorite partners to work with overall have been very coachable. Always be willing to think critically about your role in the match and how you could improve for the next one. At the higher levels of play, these small incremental improvements match over match can make a world of difference.

Obviously there are definitely reasons you shouldn’t play defense. Here are a few-

  • You spent all this time and effort on mechanical design and controls on an incredible manipulator.
    • This is entirely a valid reason. It’s unfair to everyone who worked on it to see their pride and joy just smashing other robots on the field.
  • You are the top scoring robot on the alliance
    • We only play defense when we’re reasonably confident that defense will net us the higher point differential. If it doesn’t make sense for our robot to play defense, we won’t.
  • Your alliance is outgunned and you need to do triple offense.
    • Game and alliance dependent. Use your scouting data to make this call!
    • Defense wins championships until it doesn’t.
  • Defense won’t win matches alone.
    • Your alliance needs to back it up with enough point throughput to win while effectively down a scoring robot.
    • Example: There are 6 robots on the field, A, B, C, and X, Y, Z. You(A) spend the entire match defending your target(X). This means that the combined point throughput of B and C must exceed that of Y and Z
  • The game literally doesn’t allow for it
    • But seriously, always look for opportunities to affect your opponent’s scores/opportunities for points. Can wars and noodle throwing were great examples of this.

Like it or not, defense is a part of most team games. I firmly believe that a solid knowledge of defensive maneuvers and techniques is important in all aspects of the game. Our offensive play has only improved by backing it with experience on defense.

For further watching, here are some of my recent favorite examples of defense:

  • 5930 @ Turing playoffs 2019 - Absolutely relentless on their defensive targets all bracket long
  • 498 @ Galileo playoffs 2019 - knew when to give up an inch to take the mile.
  • 5499 @ Carver semifinals 2018 - great positional defense
  • 6377 @ Newton finals 2018 - excellent rotations between scoring and contact defense

I’d be happy to answer any specific questions on the topic!

43 Likes

I want a seminar from 771 and 1073 on effective defense at champs next year, let’s make that happen.

12 Likes

We should be thanking you if anything :wink:

2016 Einstein SF1-2

5 Likes

I’ll add another reason that Jeremy indirectly touched upon.

You’re not patient enough - More often than not positional defense is the most valuable and seems the easiest. However it all falls apart when robot is willing to hold their position. There’s a huge tendency among drivers to start chasing when playing defense, but this is the exact opposite of what you’re supposed to do when playing positional defense. You need to get to a spot and hold that spot. Yes, this is boring and yes this may seem like you’re not actually doing anything. But if positional defense was the right call for the given match, by sitting there and holding your spot, you are absolute doing something, and that something is denying your opponent scoring opportunities. So you need to be patient and not drive away when your opponent does. Chances are they’re waiting for you to come off the spot, so they can take advantage of your absence.

24 Likes

This is the part wherein I mumble semi coherently about how good game design REALLY should make this strategy less effective :frowning:

21 Likes

Take a look at these few examples from 2019 of what Karthik is talking about,
2767 0:43-1:18
771 0:37-1:30
1073 0:00-2:45
What makes these teams defense so effective is that they are willing to just sit and block a position, and then once the team gets to their position, chase and poke them, and then go back into positional mode after the team has attempted to score.

5 Likes

We spent a lot of time working with 771 on the art of defense this year. It’s been a long time since I’ve worked with a team who just “got it” as well as they did. They quickly became a team that no one wanted to play against. I’m still flabbergasted that they lasted as long as they did in Alliance Selection at both Provincials and Champs.

4 Likes

Must have been a really good teacher, watching them was amazing.

1 Like

I appreciate the information and examples in your post, I have some quick questions though:

What type of drills (practice not power) should a team be running to improve defense?
Is it advisable for a robot/driver with intentions of scoring to also train for defense?
Are there off-robot ways to improve a driver’s defense mindset?

1 Like

Fantastic post and lots of really good information!

Strong tactical defense has become a priority focus of our team after our first event last year, and I think it’s safe to say it’s now become one of the things our team specializes in. I see so many teams just ramming into other robots and considering that “defense” even if said ramming is doing nothing to block the target robot from scoring or getting to their destination.

I also see a lot of teams that are almost too afraid of getting pinning penalties. A good driver can get a good 3-4 second pin and back away long enough (and fast enough - another reason quick acceleration is important!) to avoid the penalty before going back in. Maximizing the pin duration can play a huge role in how effective you are at getting in the way. As @Jeremy_Germita has already mentioned, getting your driver more stick time is probably the single best way to improve your defensive game. Bonus: you’ll help your offensive game too in the process, so start training your driver(s) as early as possible, off season if you can.

2 Likes

I know this may not be available for your team, but if you are able to have two running robots at once, this will be some of the most effective practice you can get. Have something that semi-mocks up the field, have a garbage can or something as the goal, have one driver’s job to be stop them from getting there, and the others to get there. Make sure to teach that blocking them from getting there is normally more effective than pushing them away. During this “game” be sure to have normal defense rules in effect. A driver who practices playing defense, will be better at playing around defense.

Yes.Yes.Yes.

I’m struggling to come for something for this, but I’m sure there is. If your driver can relate to sports, you could have them think of it as being an offensive lineman, and the goal is the quarterback, your job is to stop someone from getting around you for as long as possible.

2 Likes

It actually wasn’t until a recent robot demo that we started to notice the value in having multiple robots driving around. The main driver rotated between defense and offense and the driver of the other robot would change sides as well.
Nothing too scientific, but we definitely plan to do this more often in preparation for events.

I’d argue that after playing defense at our first event, the way we approached pathing and scoring completely changed for our second event.

Also it doesn’t hurt to have that defensive experience, even if it’s just practice. Never know when things go awry and you need to pull out the ol bumper to bumper.

Open to suggestions on this one, but we like watching match videos obsessively. Identify what great defenders are doing well, what good offense does in reaction to defense, figure out how to react to that, etc.

All of them. Not only should you be running drills which specifically simulate each of the defense types you expect to play, but literally any drill which improves your driver’s ability to drive your robot exactly when and where it needs to be will be time well spent on both offense and defense. Also, doing lots of different drills keeps your driver engaged and learning. If you can engineer some during-match switches in strategy, go for it. That’s how we put people on the moon.

As a “silly example”, 3946’s 2018 post-season robot, Mini-Koopa, was basically a 2018 Everybot without the bucket but with a piston to raise the intake high enough to launch a cube into the switch. Oh, and shorter in both height and length (only 4 wheels) and quite fast at short sprints because it was so light.
After several weeks of drive practice, I could see that Gavin (driver) was getting bored. I (rather bombastically) suggested that sure, Gavin could run the drill forwards, but did he know it backwards? After a few blank looks, I said like it was the most obvious thing in the world “Take ten cubes out of the exchange and build the pyramid!” After about two-tenths of a second of stunned-mullet-face, Gavin regained his composure and went to it like a craftsman. After that, he built the pyramid at least two or three times every practice. It was a great showpiece at demos, but the real difference showed up the few times we wound up playing defense, because he was by then so confident in handling the robot.

You said 1073 for the whole match, but their defense for the first 15 seconds was nonexistant tbh. Pretty disappointing!!!

(Being serious tho, 1073’s defense is like the perfect image of what it should’ve been this season)

We won Tesla by just running ourselfs into the rocket vs 1577 because we had never practiced defense. I think everyone really overcomplicated defense this year.

They got perfectly lined up to be ready the first second sandstorm was over, allowing them to do what they wanted with 1114, not sure what more you could want :wink:

1 Like

Top Reason:

  • You built a mecanum drive train.

Stop it. Build the kitbot.

-Mike

41 Likes

Not meaning to toss in a monkey wrench, but I’m currently building a kitbot with a mecanum drive train as an off-season project. The kitbot isn’t just for 6 wheel skid-steer anymore. The last seven I was part of building (including my current project) were:

  1. 2015: 14U2 H/slide drive, though it wound up just being 4W omni because we didn’t properly load the slide wheel.
  2. 2016: 14U2 with 10 overlapping wheels (4 2010 8" kit wheels in outer positions, 4 pebbletop 8" wheels in the main carpet position, and 2 4" 2015 kit wheels in an emergency position for crossing defenses, and a wedge end-plate kit to get us onto the first part of the defense.
  3. 2017: 14U3 with 4 6" wheels, chop the back third off the robot.
  4. 2018: Actually built according to 14U3 plan with 6 6" wheels. Wound up being the worst robot 3946 ever fielded, possibly including rookie year, though not because of the drive train, which was solid. The first frame break 3946 did since rookie year.
  5. 2018 post-season: An easy adaptation of the 14U3, using the 2018 practice end plates and mid-plate and the 2017 practice robot side plates. Apart from the drive train, had a two-motor intake and one pneumatic solenoid valve. IMHO, the best robot I have been part of to date.
  6. 2019 pre-season: Even smaller! A midget robot with 4" wheels, a frame perimeter of ~62.5", and a horizontally mounted battery inside the chassis. It’s totally awesome to be able to pull an FRC-class robot out of a duffel bag!
  7. 2020 pre-season: A mecanum configured like a number of 2015 robots, with a TB-micro on the rear wheels and a TB-mini mid-mounted to drive the front wheels. Because of some very specific choices, it can be run with or without a frame opening with a few minutes of work. I’m planning to install some intakes and a very light elevator to enable it to play Deep Space. The key to “a very light elevator” is recognizing that your scoring mechanism doesn’t necessarily have to be the same [weight] as your intake.

What are you hoping to learn from this project?

1 Like

3630 definitely begs to differ. They were a mecanum drive that absolutely destroyed us in our first event (Northern Lights) and our last (MN State Championship). There are teams that can absolutely ball out with a mecanum drivetrain.

With that said I think they would’ve been even better with a tank/kitbot drivetrain. I completely agree that kitbot > mecanum drive all day every day.

Petition for vendors to rebrand mecanum wheels as vectored intake wheels.