Shooters: What have you learned?

[billbo911]

Originally Posted by Ether

Control the shooter wheel speed with a bang-bang controller rather than PID.

Tip of the hat to Martin for reminding me of this.

Based on the above linked paper, I threw together the LabView "equivelent".
I chose to set it up with a counter instead of an encoder, but you can still use an encoder as the input to the counter. With this configuration, you can also use a simple Banner sensor and a piece of Retroreflective tape on the wheel if you would like. I also chose to use the 10ms. periodic loop.
With this version, you should be able to modify the loop delay without changing anything else. It will compensate for the change in interval time.

Please test this before you use it. This is built purely in code only and has not been debugged on actual hardware......yet.

[EDIT] I realized I had not made a compensation in the slew rate limiter to take into account different loop execution times. So, in the next version I'll have it added. [/EDIT]

[EDIT#2] See my next post, post #36, for the updated version of the Periodic tasks.vi. [/EDIT#2]

[rbmj]

I would say shooting for the backboard increased our shot consistency significantly.

Also, if you use a flexible back plate, make sure its braced to handle the force from the ball. Ours would lose a lot of energy as it was only held in place at a single contact point by a winch. You could see the plate buckle somewhat as the ball forced it back.

1218 had a similar back plate if i remember correctly, but it was braced by a solid assembly with multiple contact bearings so they had greater range and accuracy.

[Hawiian Cadder]

Originally Posted by Jeffy

Conveniently knowing both systems well allow me to answer.

Our shooter does not shoot the ball with a very high velocity (compared to most). The ball is compressed between a roller and a flat piece of poly carbonate. This allows the ball to be decompressed over a fairly large distance (~5 inches). So, it's not exactly an "explosive" exit like it is from a shooter with dual wheels spinning much faster than ours.

Morale of the story:
Allow the ball to compress and decompress over a distance rather than "instantly".

That is probably a more complete and better analysis than just "moar compression" I think that using relatively smaller wheels (3.875inches) also made the compression behave differently than most other shooters.

[DominickC]

What have we learned? Cannons are awesome

[CalTran]

Originally Posted by DominickC

What have we learned? Cannons are awesome

I really want to feel your robot up at Champs...

[billbo911]

The attached Periodic Tasks.vi include compensation for different loop intervals. This allows you to set the priority/frequency that you want to update the shooter motor drive. It also includes Slew Rate limiting that compensates for the interval time. Lastly, I cleaned up some data types conflicts.

[EDIT 4/13/12]Ether has made some refinements to his document on the Bang-Bang control. This latest version for LabView reflects those changes. [/EDIT 4/13/12]

[IndySam]

Originally Posted by hill

From what I have found, the top three options for having a consistent shot are:

1. Catapult

2. Extended acceleration period

3. Shoot it hard with a lot of backspin


Kellen Hill

You forgot #4 Fling-A-Pult originated by 3940 and used by 292, 829 and 217

[Kellen Hill]

Originally Posted by IndySam

You forgot #4 Fling-A-Pult originated by 3940 and used by 292, 829 and 217

What exactly differentiates a Fling-A-Pult from a catapult? It's just something I don't have experience with and haven't personally seen this season. After just watching a couple of 829's matches, it seemed like a pretty consistent shot. More details/pictures of your Fling-A-Pult would be helpful.

[Hugh Meyer]

PID in the Jaguar works well. It would consistently recover in about 0.5 seconds after a shot. See attached graph for details.

-Hugh

[1502]

We have a single axis shooter that doesn't have a consistent compression. Basically the ball enters the shooter at the 0 degree mark with no compression. At 90 degrees it has 1" and at 180 degrees it puts 2" on the ball. There is an extra strait (tangent) 2" after the half circle that holds the ball on track while it is decompressing and it definitely helps accuracy. We've never had any ball go left or right (I'm not saying anything about a vertical accuracy because we had some sensor issues). Anyway, the ball auto centers in the rails before being grabbed by the wheels. The wheels are basically polished and that really helped us not tear up balls (at 5200 rpm having a grippy wheel will tear up any ball) and they worked because of the compression.

All this being said, the amount of back spin we achieved was amazing and we could hit the very top of the backboard and still make it. During prototyping we tested from beyond the co-op bridge and made it, with about a 25% accuracy In the end it comes down to having had experience with designs you make during prototyping.

We wanted to keep our center of gravity as low as possible. So we ended up with it about 10" of the ground. Very helpful when balancing. I've seen too many teams flip their bot in the last seconds trying to balance. With that low of ground clearance you'll never flip. To show off we tested ours to about 70 degrees before tippage

[smistthegreat]

Our shooter hood has a gradual arc on it that allows the ball to decompress gradually, which seems to help consistency. Also, we always shoot from the closest to the hoop as we can get while touching the key, with the lowest rpm possible that will still allow the ball to hit the backboard. This combination of factors allowed us to have a very consistent bot this year.

[swwrobotics]

Our team was one of the very few that made a shooter with one flywheel on top of the ball causing it to have front spin. This made it super hard to score shots unless they were a perfect swoosh, because otherwise the ball would just jump right off the backboard. So definitely give yourself time to test out your ideas because they might not always work they way you think they will.

[IndySam]

Originally Posted by hill

What exactly differentiates a Fling-A-Pult from a catapult? It's just something I don't have experience with and haven't personally seen this season. After just watching a couple of 829's matches, it seemed like a pretty consistent shot. More details/pictures of your Fling-A-Pult would be helpful.

I posted a pick.

[ChristopherSD]

The condition of the balls really takes its toll on shooting accuracy.

All I want next year are durable game pieces.

[Kellen Hill]

Originally Posted by IndySam

I posted a pick.

Looks like another great way to deal with ball inconsistency. Consider it added to my list!