7492 CavBots - 2022 Build Blog

FYI this is some of the most useful testing of anything I’ve seen. I’m really excited to see how things go with the agitator wheel in there! Thanks so much for your efforts!

6 Likes

This will definitely be interesting. For all the bounce-outs I observed in the video, I think they involve some sort of interesting interaction with the bottom of the cone(where I know I stopped my simulation efforts). The reliance on that bottom geometry is indeed concerning.

2 Likes

We finished our full-scale shooter prototype last night and did some testing in our shop. Sadly, we don’t have an upper hub like that to test with, though :frowning: It did, however, show some interesting results with spin…

Without going into too many details, as we were working on the exact angle and power requirements for high goal shots from the Launch Pads, we found that the ball hitting the ceiling (16’ ceiling) would dramatically change the distance, adding up to 5’ compared to a shot with the same arc that just barely missed the ceiling. An while we don’t have a measurement for it, the speed of the ball increased enough to be visually evident after hitting the ceiling.

Which brings me to what I want to say about bouncing out. Relative to the robot, if you hit the back side of the funnel, think about how the rotational energy from the backspin impacts the result. It increases the angle of the bounce, sending it downwards towards the wheel. It also converts some of that rotational energy into directional energy, making the ball move faster. Now, consider the same thing but hitting the front side of the funnel. The rotational energy will decrease the reflection angle, bouncing you towards the other side of the funnel instead of towards the bottom. It also gets converted into directional energy, except that conversion goes the other way, effectively slowing down the ball.

With hooded shooters, the critical factor is going to be tuning your shot to hit the front half of the funnel instead of the back half. If you can do that, I think you’ll see a lot more successful shots!

Also to note: If going with a shooter for high shots, make sure you have the angle needed at low power to make a low shot as well. We very well may give up on high shots once we see how it works with the actual field, and it’s nice knowing we can “fall back” on doing low shots with our setup!

6 Likes

It seems like the wheel is bolted to the bottom of the hub without any standoff. Is that the case? I wonder if the wheel was 2 3/4 inches higher, would you get the same bounce out?

Thanks for doing these tests for us and I am jealous of your space.

1 Like

Smallish update here. I did make and install the center wheel. I kept it much simpler than AM’s/VEX’s design - it’s just a Redline on a 64:1 reduction, coaxial with the wheel itself. It’s entirely made out of parts we had lying around, plus two simple polycarb plates that we screwed down. We just plug it directly into a battery for shooter testing.

I didn’t really get as much time to play with it today as I hoped, so take all this with an even bigger grain of salt than usual… on Friday, our next meeting, I’ll try to play around with it a lot more. Anyways, what we have seen so far is not promising… it did not seem to help with the bounce-out at all, and if anything, made it worse. As you can see in this video (shot from Zone Three, about halfway between the Hub and HP station), the ball actually tended to hit the side of the wheel and get flung out rather than losing all their energy as we hoped.

This caught me by surprise, because at the beginning I thought bounce-out wouldn’t actually be that much of an issue This was largely due to this video from FIRST, released on kickoff day, showing how the hub works. That video shows 5 balls getting scored in a short period of time, none of which bounced out or even really seemed all that close. However, upon closer analysis, I think I have figured out where the discrepancies are coming from:

  • They score multiple balls, which may lead to them “cancelling out” their energy as they collide, bouncing off each other with a low velocity instead of bouncing out
  • All of the balls have very little to no backspin, some even seem to have some frontspin
  • The balls all come from a variety of angles
  • All of the balls appear to have been thrown in an odd trajectory, not what you’d expect from a robot

This all indicates to me that they must have been thrown by several people, not by a robot. In our experience, shots thrown by hand were much, much less likely to bounce out - and ironically, they seemed even less likely to bounce out once we added the wheel, whereas flywheel shots were even more likely than before. Additionally, I would be willing to bet they took multiple shots - after all, that clip wasn’t to show bounce-out, it was to show how long the balls took to get funneled out.

Of course, while especially with this wheel I feel confident that this is very close to the real thing, the possibility is still there that there is some small detail with the real deal that completely changes the way this operates, but we all have to just use our best judgement with the limited information available to us. And that’s why this is rewarding. With that said, hopefully in a few weeks teams will start getting their replica high goals from AndyMark and VEX, and can chime in with their findings.

What can we learn from this? Two things:

  • This seems to support my theory that too much backspin is bad and drastically increases the bounce-out
  • Having multiple balls in the high goal at once may help prevent them from bouncing out.

However, I would not count on the latter, because it is largely out of your team’s control, and the odds are VERY low that any non-playoff matches, especially at earlier events, will have any more than one or two balls in the funnel at a time. For now, my team will be adding a top wheel to attempt to reduce backspin.

25 Likes

Thanks again!

It seems to me like your wheel is spinning much faster than the wheel in the FIRST video. To they list specs on that? If so have you checked them?

1 Like

As far as I could tell they did not list any specs. It’s actually geared slower than the one AM sells - that’s a CIM on a 12:1 reduction, which is 444 RPM, wheras this one is a RedLine on a 64:1 reduction, 328 RPM. It does seem fast, though… if someone knows what the RPM speed is supposed to be, let me know and I’ll try my best to match it.

1 Like

The AndyMark one is correct according to the field drawings (page 199). I don’t see anything indicating how it is wired / what speed it is at. Might have to Q&A that.

1 Like

Yeah, probably worth a Q&A, seems reasonable. I’ll talk to my LM and try to submit one tomorrow.

2 Likes

No need. I just checked the Q&A and it is already asked, but not yet answered.

The videos I watched linked above show your upper hub being pretty rigid as balls hit it. The field tour videos shower the upper hub flexing and absorbing a good chunk of the energy from launched balls. I wonder what difference this makes in your tests.

5 Likes

I thought about that. It’s hard to say, though; the material is the same (1/8” polycarb), and the geometry is the same, so any significant difference in flex would be somewhat surprising to me. Our upper hub isn’t that rigid either, although I won’t say it’s not more rigid than the real thing, I just don’t know. It’s hard for me to tell if there’s a a difference, because I haven’t seen a good video of the field itself that shows that, just the one I linked above.

1 Like

Looking over one second (1:03-1:04) of the HUB video it looks like the belt made 6 rotations. That means the the approximate speed it was running was ~360RPM which is in line with what you are doing.

I don’t think @Andrew_L was talking about the material flexing. He was noticing that the hub in the videos wobbles significantly when the balls hit it. Yours doesn’t wobble (as I would assume it should). So now we are down to how much the field setup crew tightens down the bolts on the upper hub…this is exhausting.

1 Like

The upper hub as specified in the field drawings is mostly HDPE and polycarb, which is going to be less rigid than the setup shown. I guess the wait continues until teams have actual upper hubs to video shots on…

4 Likes

Looking at the shots that bounce out and comparing them to the shots that don’t I noticed the shots that bounce out often hit near the center of the goal and come in at a steep angle/close to vertical.


The shots that don’t bounce out often look to to hit higher up in the goal and hit at a shallower angle. I think this might be because the further down the ball travels the more it accelerates thus more kinetic energy meaning when it hits the bottom of the goal it bounces up a lot. I realize

Example of shots

image
Score (hits higher without as much energy and at a less steep angle)
image
Near miss (it hits at a steep angle and bounces off the ball at bottom)
image
Miss (same issue as last shot)
image
Goal (bounces off the rim so does not have as much energy as it travels down so it doesn’t bounce as high)

Watching the Hub field tour video the balls seem to hit the goal at a higher angle and bounce from one side to the other of the goal.

I realized this watching the WCP CC the balls seem to go in because the robot is shooting the ball at the far side of the goal and not at the bottom thus the ball bounces from one side to the other.

I think the ideal shooting scenario is hitting the far end of the goal at a shallower angle like this.

non ideal shooting scenarios are hitting the goal at a steep angle causing the ball to bounce up and out or hitting the edges at a steep angle and bouncing down into the goal causing it to bounce out.


Other things that affect shots are multiple balls hitting each other, hitting the rim.

This is just my observations of the videos. Our team still cannot meet in person so we have not been able to test high goal shooters.

30 Likes

Yep, this checks out. We thought at first that hitting it off to the side would be better so it would sqirl around the goal and go in, but this is definitely more repeatable. We’re finding the strategy to be to minimize the amount of energy the ball has as it enters the cone, so we are trying to clear the cone with as little vertical height as possible (lowest hood angle and shooter wheel RPM), and then reduce backspin with a hood roller. When we implemented all of those, bounce-out was greatly reduced - at that point it only would bounce out if it hit the wheel directly (so maybe 1 in 6 shots? Tuning could probably get you even farther). I uploaded more videos of our testing, mostly focusing on the affects of the center wheel on the balls. If you would like to do farther analysis, there are quite a few videos of us zoomed in on the cone to allow us to see the behavior of the balls better.

Right now, we’re working on building the overall robot itself. We’re on schedule to finish that by the end of the week. As-is we think we’ve learned all that we need to know in the way of shooter geometry from the prototype, so we’re just going to run with it now. I’ll make a post on our progress tomorrow. Our big shipment of cargo (20 red, 20 blue) also shipped a few hours ago so I’ll do some testing on that to see how consistent the balls are, in terms of shooting (both inflated and deflated), wear, and deflation over time, so I’ll continue to update y’all with our findings.

As always, thanks for reading, and let me know if there’s anything in particular you’d like us to test. Till next time,

Ethan Ocker

10 Likes

First post in a while that’s mainly focused on the bot. I’ve been focusing on shooting balls, at least in the blog, because I think that’s what is most useful for other teams. However, behind the scenes we’ve been making a lot of progress on the bot.

First - our schedule

The schedule planned for us to finish the robot mechanically by this Thursday, the 27th. We knew this was ambitious, unreasonable even, but we thought by putting an aggressive schedule in, we could avoid the “eh, we have 6 weeks left, it’s fine” rut we found ourselves getting caught by in years past. Because, ultimately, 8 weeks until our first competition doesn’t mean 8 weeks to build a robot. We did, however, still give ourselves two weeks of “float” between the mandatory 2 weeks of driver practice before our first comp and the week we set aside for programming to get everything figured out. We knew we would lose days in all parts of it, but as long as we kept working diligently, we would be fine. And, well, seems like we’re almost there; we won’t mechanically finish 100% on schedule, but we can realistically get everything done by the end of this week, which would be huge for us - not only is it the most complex and ambitious and complex robot our team has ever done, it would be the fastest build we have done. By far.


Anyways, as for the smaller details. The overall CAD is almost done, it’s just a few things that have been worked out but need to be modeled. The only things left are the intake mounting (we’ll switch from the two 1x2s to flat plates that bolt to the upright 1x1s), the climber (we’ve already figured all the details for that out - TTB 2 stage climb, powered by a 12:1 or 15:1 MAX planetary), as well as the control system and electronics mountings.

As for individual mechanisms, we’ve finished machining for the entire shooter, including the hood and turret, and have made progress in the rest of the mechanisms. Today marked something of a milestone with our first subassembly completed, the turret. Ironically. That was done first not because it was the most important, but because our team had never built a turret so, if we deemed it unreasonable, we needed to know right away so the rest of the designs could be changed accordingly. Fortunately, we found something we’re happy with, so we rolled with it.

Also did some stress testing by screwing it across two tables, similar to how it will be mounted on the bot, and then standing on the mounting points for the shooter. Idea being that if it could hold >300lbs of person standing still, it can hold <30lbs of shooter moving around. Scientific? Not really. Safe? Probably not. But hey, it held up!

Short term next steps are to finish the shooter.


We have all of the parts here, at least as far as we know, so with any luck I’d like to keep our momentum going and finish assembling that tomorrow.

Small learning experience here: I told the team (mostly directed at the CAD/design people) that, with how shipping is these days, if their design requires a part that is out of stock, backordered, or relies on something that ships out of the country, it was a no-go and they had to redesign. We couldn’t risk our season on a random part that got stuck for a week. Fortunately, we’ve been alright, but especially with the SDS Mk4s we ordered, we can feel the impact the shipping delays are having. One example of how this rule came into play was the WCP Nut Strips - the shooter required two of those, but they were backordered with no clear signs of when they would be in stock, so instead we overnighted some .5x.5x12" 6061 bars from McMaster-Carr and made them ourselves. Worked perfectly, and we were able to move forward with the assembly.

21 Likes

This is really cool. I am looking forward to seeing it in-person at the Channelview event.

End of Week Three update
Pretty long post here. It covers our progress in the season thus far, our next steps, including what needs to be done if we want to add a High or Traversal climber, and then at the end going into more detail about the mechanisms that have been completed but not yet covered in this blog.

Yesterday we pulled a long meeting (from 8am to 5pm) and got a LOT done. We’re coming very close to being able to officially hand off the robot to electrical to do the final wiring. Here is what the robot and schedule looks like so far:


What we have gotten done:

  1. All machining
  • Minimizing the amount of aluminum that we needed to machine, mostly by using VersaFrame, saved us a huge amount of time here - we only have two machined tubes (which are identical), and all of the parts for the turret were cut out of a single .19" plate. The rest of this is polycarb or HDPE, which we can machine much, much faster than aluminum.
  1. Assembly of the intake, shooter, hopper, and chassis
  • We opted to use almost exclusively belts here. I have mixed feelings about whether this was a good call - while belts are very straightforward if you get them right, this proved to be a struggle because we forgot to order or had the incorrect sizes for several of them. Inevitably, considering the sheer amount of belts (almost 20 unique sizes), it’s not surprising. We made it work though, just needed to use oversize belts and idler shafts to tension them properly. Lesson learned for bot 2.
  • Other than belts, we are using Ultraplanetaries/Maxplanetaries for all of our reductions - these worked out great. Much cheaper, more compact, and lighter than a spur gearbox.
  1. Mounting the turret, shooter, and hopper to the chassis

What we still need to do:

  1. Mount the intake to the chassis
  2. Assemble the climbers and mount those
  3. Finish the “shooter shelf” that holds the Igus chain and most of the control system
  4. Do all of the wiring

The hope is to finish this first robot and hand it over to programming by the end of this week. The important thing is to keep moving; the robot is by far the most complicated one we have ever done, so we can’t really afford to take our foot off the gas.

Next steps

After the first bot is finished and running, the next major obstacle is to assemble the second bot, the one that will actually go to competition. This should be faster, because in the process of the first one, we’ve been taking notes on how to change the design, etc to make the second one easier to assemble and manufacture, and of course the design is already complete for it.

We’re also looking at adding a L3/4 climber. Our current plan for that involves a third telescoping arm in the center of the robot, which can rotate at the bottom. Since we don’t have pneumatics, this requires two motors, and as-is… we have no motor slots remaining.


We have 19 motors currently, with the last slot being taken up for our CUSTOM CIRCUIT to power the Limelight, sensors, etc. So, if we want to install the third telescope, we have to save two motors somewhere. As we see it, there are a few ways we can do this:

  • We can use pneumatics. This could save us a motor to raise/lower the intake, and potentially also work for rotating the L3/4 climber.
  • We could reduce the amount of the motors being used - we’re waiting to finish the bot and do some testing before making a decision here, but likely we could do a single motor for the shooter and not power the hopper walls at all. Alternatively, we could make the intake not be able to rotate back up at all and simply remain locked down. We are hesitant to do this because the intake is far above the height for bumpers, so we could easily hit another robot within it’s FP if we are not careful.

Of course, there’s other issues with a high climb that would need to be address before we could run this in competition

  • Since we are using swerve, a fall directly onto a module could be potentially catastrophic, so it has to be fast enough to be worthwhile while still being incredibly reliable. This means even more work for programming, on top of the already highly complex shooter.
  • Mechanically, we’re not even sure if it would be within weight - our robot weighs 95lbs according to the CAD, but that doesn’t include any of the electronics or wiring, which would likely weigh as much as 20lbs by themselves (8lbs for the 11 Smaxes, 4 for the rest of the CS and sensors, and then 6lbs for the wires and energy chain), which brings us to 115lbs. The telescoping climb itself would weigh about 8lbs, excluding all the reinforcement that would be necessary for the climb to bear the entire weight of the robot as it swings from bar to bar. We’ll just have to see.

Finally, the mechanisms. Yesterday we fully completed the hopper, and got that mounted to the chassis, allowing us to mount the shooter on top. We also came close to finishing the intake, and the only things we really have left are just doing the superstructure so we can mount it.

Starting with the hopper.


It consists of two main subassemblies; the hopper floor and tube, driven by a NEO on a 12:58 gear reduction. This brings the balls into a single file line and then throws them upwards through the turret into the shooter assembly. The walls are powered by a N550 on a 10.2:1 reduction (4:1 and 3:1 UP), with a long belt connecting the two sides together. We expect we can safely make the walls unpowered and still work fine, but we are waiting until we can do some real testing to make a decision here.

Next step was mounting the shooter/turret.


Had a good time using clamps to hold everything in place - it had to be centered to the hopper, and the tubes were slightly out of parallel, so using almost all the clamps we owned got us there.

Last but not least, the intake.
image
It consists of 3 rollers with TTB squish wheels every 2", and then will have a “flapper wheel” in the front to catch bouncing balls and direct them into the intake. The whole thing is powered by a NEO on a 14:24 belt reduction, and will rotate up and down with a NEO in a 15:1 MP, then 18: 30 and 18:42 belt reductions. We chose to use motors as opposed to pneumatics to rotate it because at the time of designing, it was the only part of the robot that would be using pneumatics. The entire system would have come out to something like 8lbs (iirc), as opposed to about 4lbs for the motor and gearbox.

That’s all we’ve got for this week! Thanks for reading and let me know if are any specific things y’all would like more details on, or any questions you may have.
Till next time,
Ethan Ocker

11 Likes

Big milestone here! On Monday, we finally got the bot to a point where we can hand it over to electrical. That’s not to say mechanical is done, there’s still a decent amount of work that needs to be done to get the first bot in full working order. Realistically, no one’s job is ever really done. But, this is still the fastest we have ever gotten a robot to this point, which is especially exciting considering that this is also by far the most ambitious and complex robot we have ever done.



Electrical is hoping to finish all the wiring by this weekend, and then we scheduled a few more days so that we could troubleshoot any issues with mechanical/electrical. Goal is to have a fully functional bot by early next week, at which point mechanical/electrical can put their full focus on the second bot (we don’t want to sink resources into a copy of a design if it’s flawed to begin with), and then programming will have primary access to it for a few more days to get everything working smoothly and have some time to work on autos.
After that, it’s the driver’s turn, and hopefully the second bot will be done shortly after that point so programming can focus on the more high-level things (such as shooting while moving), and working on getting autos working. Once both bots are running, our next priority (depending on time) will be getting a L3/4 climb.

Unrelated: this year we’ve been using a lot of REV’s new MAXplanetaries. They’re great, we’re using 2 of them for climb, one for rotating the intake up and down, and then one for the turret. Yesterday we were testing all the individual mechanisms and realized that the intake rotation (which we had not yet tested) was completely locked up, so we disassembled the planetary and found that the 3:1 stage was defective. As you can see in the picture below, one of the 3:1 center gears was made incorrectly, which lead to it sticking into the spline that is supposed to connect one stage to the next, therefore seizing it up. We have an extra and are going to reach out to REV, this is just a PSA to always make sure you check that everything runs smoothly before committing it to the final robot.

That’s all I have for you today. Hopefully next post will be a “completed” robot. Almost there! Can’t wait!

12 Likes