Catapult?
 

I haven't seen any discussion on anybody using a catapult design - is anybody going this route? Has anybody tested this? If so, is it accurate and effective? I imagine this design would be very difficult to defend (aside from breaking the arm off). Just curious....

Re: Catapult?
 

I can't speak for everybody, but team 1318 dismissed the catapult idea early in our first brainstorming meeting as hard to reload, hard to make accurate, hard to shield, and hard to make as powerful as necessary. I will be very impressed with any team that pulls this off successfully.

And I'm not sure what you mean by "hard to defend". I don't see how it would be harder to defend against a catapult design than any other type. And of course, breaking the arm off would be against the rules, so that isn't a legitimate strategy.

Re: Catapult?
 

I think a well-constructed catapult is likely to be more consistent in its aim than a wheel-driven shooter. I also think it's pretty obvious that a catapult will be slower than a wheel-driven shooter.

Any teams designing a catapult will have to be careful not to run afoul of the very first rule.

Re: Catapult?
 

A trebuchet would potentialy work (especially a Floating Arm Trebuchet) and would still be considered safe by <S01> regarding the large mass being stopped abrubptly, however, a large arm, and a large mass would be required for a good trebuchet, both of which would be difficult to acheive with the game limitations...

-Leav

Re: Catapult?
 

I don't think a catapult is a good idea. If any team can build a successful one, then by all means, go ahead and try...but it will indeed be difficult to work with..Unless it can scoop and shoot.

Re: Catapult?
 

I dont know, a catapult for this application would not have to be very big - maybe a 10 or 12" throw

esp when it can be actuated with a pnuematic cylinder that can provide up to 180 lbs of force (2" diameter cylinder charged with 60psi)

Re: Catapult?
 

Actually, my spring tensioning cam was stolen from a catapult of sorts. I don't know the name and thus can't google it, but it was designed to throw basketballs. It's a little hard to explain the exact mechanism, but effectively, it was a lever with springs on one side of the fulcrum and a ball holder on the other. The thrower has super heavy duty springs (possibly garage door springs) so that it has a short throwing arm.

With the basketballs, it was able to throw about 30' and hit the same spot within a 1' radius. Personally, I think a similar device would be much more repeatable than most other ball throwers that are being discussed in the forums. For one thing, nothing is moving just before firing. This ensures that you have the same conditions every time.

Re: Catapult?
 

i also dont think that a catapult is a good idea. it may be accurate but it takes even more to get 1 ball to fall into the catapult at once, unless you want to shoot more than one at a time

Re: Catapult?
 

Not to toot my own horn, but two or three years ago, when we had the last non-autonomous period, we had a game based on soccerballs. We built a catapult powered by constant force springs (the springs made of coiled sheet metal) using 6 5lb springs, we got our robot to shoot soccer balls 99% effectively. The good thing about this was that our robot did not shoot at different velocities because of the change in our battery level.

Re: Catapult?
 

Our first design was a modified trebuchet.
But it was 1500 kg and was over 8 meters tall.

Next we built a giant rabbit...and we crawled inside it waiting for the right moment after autonomous mode...

Re: Catapult?
 

Now I'd *love* to see how someone designs a Trebuchet "repeater".
Resetting it to make the SECOND shot (esp autonomously) is
NOT a trivial robotic design problem.

But even if you *could* solve that one, it'd have to be a TINY
Trebuchet to comply with this year's build shooter sizing rules.
IMO, your biggest problem is getting it into the original volume,
to comply with shooter sizing rule <S03>:

> <S03> Shooter Mechanism must remain inside the ROBOT
> - Any mechanism used to throw balls must be contained within
> the original 28” x 38” x 60”starting envelope of the ROBOT [...]

Hmmm... How would a catapult shooter deal with this rule?

- Keith

Re: Catapult?
 

how much 'throw' do you think a catapult will need to launch the nerfball at 25mph?

we are not talking about launching a car 2,000 yards through the air here :^)

Re: Catapult?
 

Well, let's take a crack at that one...

Furthest case: At the end of the stroke, the center of the ball needs
to be moving at 12 m/s max. [edit] IOW, a point on the circle made by
the ball cradle's sweep that intersects the center of the ball must
be traveling at 12 m/s. That means the final rotational speed will be
related to the Throw Radius, which I'll define as distance between
the pivot center and the ball's center. [/edit]

We've analyzed 48 Poof-Slinky balls. Amazingly, their sample
weights are fairly evenly distributed between 175 and 186 grams,
which places the median at about 180-181g. Therefore, we then
calculated a ball will contain roughly 13.7-13.8 J of energy at 12 m/s.
(Can someone else here, please check our math? Are we right?
Anyone have more weight samples, to improve our data set?
We suspect our sample set is too small, because we didn't
get the expected 'Bell Curve distribution'...)

[edit] To calculate the ENERGY required, you now require: the
arm's parameters (specifically the arc range of motion, the
distance between the pivot and the center of the ball, and
the rotational inertia of the arm). [/edit] From those, it should be
possible to calculate the torque required to accelerate the ball
over that arc to impart the required energy, and guarantee the
system still meets the "12 m/s max muzzle velocity" FIRST spec.

You then check to see if you have a motor still available capable
of imparting the required torque, and design an arm torque system
that'll properly clutch itself to prevent self-destruction from
repeated "stop shocks"... :)

BTW, a big question for all: Is there an easy way to hand
calculate the rotational inertial of a wheel and/or arm design,
to evaluate them before construction? Weighing it is insufficient,
as you need the radial mass distribution.

Professionally, the only way I know of is to Solid Model it with
mass properties on, and run it through a Mechanical Simulation
System (MSS) software suite like ADAMS. But we don't have
that software here now. It'd be VERY expensive (and take awhile
for approvals) to buy it for the school, and we don't have training
time during the build to learn it, even if we could get ahold of
a seat or two of it... (Darn...)

We also don't have the time to build a bunch for testing. Otherwise,
the other "simple" way would be to "make one, apply a known torque,
and measure its acceleration" with some improvised measurement
widgetry (like the time it takes it to traverse a known arc between
two retroreflectors). Then tweak, and repeat... <sigh>

How do you simply measure (or calculate) the rotational
inertia of an arm system, or of a sample object (like a wheel
you have on hand), before building up a system using it?

Back to catapults...

Now a problem with catapults in general is that (when compared
to continuous feed methods like belts or flywheels) unless you
"complete the arc full circle", you are wasting a LOT of energy
per throw. With a traditional arm catapult, you have to put in
the energy to accelerate the ARM as well as the ball, and the
remainder is all lost when it hits its stop. Since the arm is
normally a lot heavier than the ball, the relative energy cost
is huge to launch a bunch of balls. And, you still have the whole
"recocking cycle time" issue to contend with.

OTOH, if you think a set of "Jai Alai" style shooters (Cestas) arranged
like spokes around a wheel, you may have something there! :)
You KEEP the energy you spent revving up the wheel, and now
only have to replace the momentum lost by throwing each ball.
You either "drip in" balls, or catch them off a dispenser to throw.

Now getting this all synced up to have the Pelota (Spanish
for a Jai Alai "game ball") properly intersect the Cesta
(Spanish for a "basket", the Jai Alai "shooter") at the right
point in each cycle to make it repeatable would IMO be an
interesting technical challenge.

Controlling exit angle is another problem. A combination of where
it first intersects the Cesta and its rotational speed determine
how long it remains on the Cesta accelerating. So, exit angle
COULD be controlled by adjusting the intersect or radial drip point,
(or by changing the wheel's speed), but all this could require a LOT
of experimentation and tweaking!

Of course, assuming that all worked, how do you then "shield"
the whole assembly to match FIRST's safety rules?

I think other rotary designs may be easier to package,
aim, and be more repeatable, which is your true desire
here. (You need repeatability first, before you can possibly
calibrate your shooter.)

But man oh man, it'd be REALLY COOL to see a two to
four armed "Cesta Wheel" (Cesta Rueda? Basket Wheel?)
in action, whipping out balls! :D

Comments?

For more info on Jai Alai, some terminology, and to see
a real Cesta (a traditionally made wicker basket shooter),
Google it, or go to: http://www.dania-jai-alai.com/page6.htm

- Keith

Re: Catapult?
 

I think you have made some assumptions that are not necessarily true.

1. whatever your catapult 'arm' is made of, once a ball is fired you must stop it before you can fire again, so the inertia of the arm is always lost (unless you are going to have something more like a bat that wacks balls that are dropped in front of it.

2. an arm that rotates only part of a circle, 90° for example, could hit a spring at the end of its motion that bounces it back into its starting position. If fact, when it gets back to its starting position there could be a second spring that absorbs (recovers) the momentum of the arm?

3. a motor would be my last choice for transferring energy into a catapult. First choice would be a pneumatic cylinder.

Re: Catapult?
 

Actually, I stated that as a disadvantage. (Sorry if I wasn't clear,
but I mentioned about the energy lost with a "traditional arm"
design.)

BTW, in FIRST contests (IMHO) "continuous operation widgets"
are always superior than any oscillating "batch" device,
The total average cycle time per item handled is much shorter
(especially if the batch device only holds ONE field object, and
there are many you need to process in a round to win)...

That's why I mentioned a "Cesta Wheel" ([TM] ;) ) as a possible
"catapult variant".

Hmmm.. It sounds cool, but I love to see that work in practice,
without risking tipping over the machine. I believe the energies
and lever arms involved are high enough compared to the total
machine mass (and its distibution) to be of concern. In a
traditional catapult, the base to arm & rock mass ratio is pretty
high. Now it may not be a problem, but without actually crunching
some numbers on a proposed design to make me feel better, I'd
worry a lot about the possibility of the machine being kicked over,
or at least having its aim ruined for the next shot. It wouldn't be
good to have to "set up" every shot with a move.

Well, in this application, I'm not so sure...

Let's consider cylinders for a moment. You can get some pretty
strong forces (up to 188 lbs directly), and you can stall it forever.
That's cool.

But a big cylinder is SLOW. I have no clue how you'd fire a lot
of balls with it in a short period of time.

Also, you don't have much total work available in a FIRST
pneumatics system, when compared to the motor/battery
system. The air storage is pretty small. The pump delivers
only a small amount of air in 2 minutes. And, you're limited
by the small hoses, causing the cylinder's slew rate limit
to be of concern when looking at cycle time.

This is a fast access, high input "total energy / round" application.
To be considered a real shooting machine, you'll want balls
to <wham> <wham> <wham> out of this thing! For Auto
mode, it should cycle faster than 1 ball/sec, and that's
doesn't even include any drivetrain positioning time to
set up the shot.

So if you're serious about shooting a lot of balls (vs just one
or two during the round for show), I believe the sheer number
of balls you want to process in 2:15 probably requires a lot
more total energy than the pneumatic charging system can
possibly provide to you, both totally and incrementally.
The cylinder slew rate limit alone suggests you should charge
a fast release energy storage device (e.g."a spring").

To ME, given that (and FIRST build rules limiting what
actuators you're allowed), it implies our choices are between:

- A BIG coil spring and an incremental release system
(kinda dangerous to work with);
- A direct motor drive; or
- Some sort of spring & motor, or spring & cylinder
combination, where the spring is recharged by either
device (and I'm still not convinced a cylinder can keep up).

However, you've raised a VERY valid point about energy recovery.
Whether you're using a cylinder, a spring, or a motor, recovering
least some of the energy is a good idea.

How about using a spring and a ratchet to fire the arm?
You "<recharge> a spring" with it, via a ratchet. (Hmmm...
CDF won't let me use the "c" word... :)) That way, IF
you can recover any energy with a "kickback spring" as
you suggest, it's stored with a <click> <click> back on
the ratchet when the arm slams home. The remaining required
Potential Energy is then added in via a cylinder stroke.
or a motor/gear combination (no solenoids allowed yet...).


Now, let's consider MOTORS. Motors are by definition a
torque device. Near stall, they put out a LOT. And, you
have quite a bit of total work energy available in the battery,
which can be tapped fairly quickly.

With a motor, one option would be "one way drive": Use
a motor/gear combination and a ratchet (or a non-backdrivable
gearbox) to <set> a BIG spring. With this system, you'll need
some kind of a triggerable clutch to unload the motor/gearbox
and release the arm. That catch could be run by either a
small cylinder, a small motor, or a servo. Once shot, re-engage
the motor/gearbox, and "wind 'er up again". (Your release is
a potential safety hazard and wearout item, so design it WELL,
with safety "lockout pins" for pit work, etc...)

The other way is to DRIVE the arm BOTH WAYS, with a
high torque, continuously connected motor/gear system.
Use mechanical or air springs on the "shot end" to help
decelerate the arm, and recover a bit of energy for
starting the return motion. Having some opto switches
or a pot to monitor the arm allows you to auto sequence
the shot with software, and eliminates "end of travel"
stall currents. It would also allow software to program
the entire cycle as a "semi-automatic or full-automatic"
shooter. But with this one, you're going to want some
kind of over-torque clutch or shock coupler, to help
manage stresses on the motor/gear system.

[edit]
Actually, I just thought about one more option: A direct
motor/gearbox, and a spring acceleration startup assist.
Now that spring could be recharged with a cylinder!
But, now you're talking about both a pneumatics package
and a big motor system. Weight might become a problem.
[/edit]


I hope this helps!

- Keith