FRC488 2012 -- Gojira
 

I wish I had better pictures or more video to offer right now, but I wanted to get something up here in advance of our competition next week in Texas and our practice robot is a sad state of affairs right now.

Existing video: youtube.com/teamxbot

These were taken as the robot was being crated up, so they're not the most telling photographs.



The above shows the right side, with the collector on one end (toward the right in this photo), our ball storage system and elevator on the near side, and catapult in the middle.



This is the left side; nothing very exciting going on here, but you might be able to get a better look at the catapult.

Technical Stuff:

Drive--
- 4 CIM motors through AM Super Shifters power 8 6" AM Plaction Wheels with roughtop tread
- Servo shift
- Passes over barrier with the aid of HDPE skids at the front and rear (missing in these photos because they were in the way of the shipping straps)
- Delrin wheel pod plates
- Pneumatic brakes

Collector--
- Driven by 1 RS-775-18 motor through a CIM-U-Lator gearbox
- Uses polycord belting to drive the intake roller
- Intake roller has some spiraled surgical tubing on it to encourage balls to move to one side of the collector
- Can store no more than there balls in the collector at a time; if a fourth comes in, it knocks out one of the others
- Raised and lowered by two 1.5" bore pneumatic cylinders so that it...
- Can lower the bridge; this was tested on a properly weighted bridge
- Can raise a horizontal bridge, lowering the opposite side
- Can stop a bridge from rotating beyond horizontal by acting as a stop below the bridge
- Dumps balls into storage when its raised back into the robot

Storage--
- Can store three balls inside the robot for travel
- Is hard to see in pictures because it's made almost entirely of polycarbonate

Elevator--
- Uses a small bore, long stroke cylinder to raise one ball at a time into catapult cup
- Has state-of-the-art rubber band system in place for keeping balls from rolling below it when its raised and jamming the works
- Needs a little more attention than it's received so far, but it works well enough for now

Catapult--
- Uses camera/vision processing software magic to align the chassis with the hoop and find the distance to the backboard.
- Automatically adjusts firing range and arc based on distance
- Can score from the fender to about 25' away
- Uses two 2" bore, short stroke cylinders and a simple linkage to fire
- Uses 5 solenoid valves and a servo-driven ball valve to do all of the above



The above is a schematic of how our pneumatic system is laid out. Yeah.

So, that's it. I wish I had some more interesting pictures to share, but these'll do for now.

We're competing in San Antonio, TX next weekend and in Seattle during week 4. Ask questions if you've got 'em! We're really happy with our finished product this season -- it's been very effective in testing and practice and we're happy that it's not just another wheeled shooter -- though we did prototype one of those.

Re: FRC488 2012 -- Gojira
 

Looks like another great machine from Xbot! I'm excited to see you guys in action at Seattle!

Re: FRC488 2012 -- Gojira
 

Any reason for servo shifting?

Re: FRC488 2012 -- Gojira
 

Did you look at the pneumatics diagram? :) We were worried about air consumption.

Re: FRC488 2012 -- Gojira
 

I'm sorry, I missed it somehow. :o Great robot!

Re: FRC488 2012 -- Gojira
 

No worries :) We have a lot of pneumatics on board -- our entire ball handling/shooting system relies on air to work, so we wanted to do everything we could to conserve its usage where we could. The servos seems to work just fine for shifting for us, so it was a no brainer to go that route and save just a bit more air for the catapult.

Re: FRC488 2012 -- Gojira
 

Here's a clearer pneumatic schematic.

I'm very interested in learning more about how other catapult robots regulate distance and/or arc since that was our main design challenge this season.

The system is divided into four 'zones' that each have a different working pressure.

The first zone is the high pressure storage. We have 90 cu. in. at 120 psi and are using a Rookie Kit compressor. It is highlighted in green.

The high pressure zone is separated from the remaining three zones by a regulator as required by the rules. The second zone, therefore, is the 60psi zone. This is a small zone -- really just a bridge between the remaining two zones and is orange.

The "everything else" zone is further regulated down to 30psi to conserve air and is everything at the bottom left of the schematic. This is where our pressure release valve is located and is red. This is where most of our robot functions live -- the collector deploy/lift, the collector flap rotation (for fitting into the bounding box), the elevator that loads our catapult and the brakes.

The final zone - in blue - is the catapult operation.
The firing sequence is as follows:

• Use camera to calculate distance to hoop and determine appropriate catapult settings
• Open single solenoid valve from 60psi (Orange) zone to Catapult Zone and fill until Fire Pressure Sensor reads expected value, then close the valve
• If actual value is greater than expected value, open Leak Flow Control solenoid valve to relieve pressure until expected pressure achieved.
• Repeat above two steps until system is within expected pressure tolerance -- about .5 psi
• Simultaneously, a servo is adjusting how far open the Outflow Valve is.
• Open valves to Catapult 1 and Catapult 2 cylinders
• Ball shoots, it scores! woo.
• Close Outflow Valve
• Open single solenoid valve at bottom of diagram to push air from 30psi into Catapult 1 and Catapult 2 to retract catapult arm
• Close that valve
• Ready to fire again.
  
  I think that's it. There's a lot going on there, but it only takes two or three seconds for the entire process and we're working on getting that down lower -- it's just a matter of tuning the flow control on the leak pressure valve to make hitting our target happen faster.
  
  

Re: FRC488 2012 -- Gojira
 

All the pneumatics rules seem to be satisfied. I'm definitely impressed, it is much more complex than any system I have ever built*. You might be able to cut down on time by using a servo actuated (relieving?) regulator in between the orange and blue parts of the system just after the solenoid. This would completely eliminate the need to have a bleed valve hooked up to a flow control which seems to be the bottleneck in efficiency. This would also conserve pressure since you are no longer bleeding away air. A good regulator and servo should be precise enough to be simple to program.






*Mostly because I have not had a chance to build this, yet.

Re: FRC488 2012 -- Gojira
 

I think I just got more ideas for how to use pneumatics in future robots that I have in the past 2-3 years. Thank you very much for writing this up and I can't wait to see this on Thursday.

P.S. I think your Double/Single labeling in the red zone is flipped.

Re: FRC488 2012 -- Gojira
 

We also tested a system with a servo-driven regulator, but were unhappy with the reliability. We waste a bit more air this way, but we're able to get very consistent firing results from it as a consequence, so it's an okay trade.



Come by our pit and see it. I'm not sure what state of affairs things with the robot are currently in since I've been buried at work and haven't made it to a few meetings this season, but I'm sure you can grab someone to give you the guided tour if you're so inclined.

Also, I'll see about getting our pneumatics mentor to post more about the solenoid choices; he says they're not mislabeled, but that's all I know about that :)

Re: FRC488 2012 -- Gojira
 

Thaaaat's a fancy pneumatic system. Very impressive. The only thing I can think of that would make it more impressive an responsive would be PWM control of your catapult solenoid valves. If you have fast enough solenoid valves, you can actually send a PWM signal to them to modulate the flow rate. So 100% duty cycle when you're first loading up the pressure, then you reduce the duty cycle as you approach your setpoint. If you do it right, you hit your setpoint and don't need any bleed-off. You could PWM the bleeder as well, obviously.

Granted, if you get your tuning wrong, you'll overshoot just as bad as you are now and take longer doing it. Or restrict things too early and take forever getting to your setpoint. But the plus side is you might not need any new hardware to implement it, just fancier programming.

Re: FRC488 2012 -- Gojira
 

About the Single/Double solenoids, this is something that is often confused, and I think I've gotten it figured out how best to explain it.

The solenoids control a valve that will let air come out of port A or port B. The solenoids should be called "Single with Spring" and Double. The valve has 2 positions it can be in, and the solenoids (and spring) are what switch the valves positions.

Starting with the Double, each solenoid is going to push the valve to one of the two positions, so we'll label them A and B for which port they push to. If you fire solenoid A, air will come out of port A. Once the valve has been pushed into one of the positions, the only way to change it is to fire the other solenoid (assuming you have removed power from the first). So the valve will stay in the last position you left it in (critical for our brakes to remain... broke? braked? after the match has ended and we lose power)

So the difference with the Single is that you only have one solenoid. There is a default position, lets say A for this. So the spring is always pushing the valve to A. The solenoid thus is what will change the valve to output to B. The difference here is that you must continue applying power to hold it there. As soon as you let go of the power, the spring will move it back to the default. The Double just needs a quick amount of energy to switch, and then you can stop powering the solenoid.

So the confusion often happens when you see varying amount of hosing coming out of solenoids. This is because the hosing come out of the valve portion. These are technically called solenoid valves, so it has two parts to it. All of our valves (and likely 90%+ of all teams) are 5 port 2 position valves. That maps to 1 input (P), 2 outputs (A and B), and 2 exhausts (EA and EB).

There are different types of valves you can use, but this does not change how the Single and Double solenoids work.

So the fact that the one solenoid we have marked as 2 only has one hose coming out of it is a result of the fact that we're using pistons that have a spring return, thus we don't need to power output B to retract them, and is independent of the type of solenoid and valve that we have there.

I hope that explains everything sufficiently

-Ben

Re: FRC488 2012 -- Gojira
 

This is just my first impression but it doesn't seem as though you've taken the KISS route ;)

Interesting design to say the least, that drive train looks solid.

Good luck Xbot!

Re: FRC488 2012 -- Gojira
 

Actually, it looks pretty KISS to me... but then again we used something similar on our ball "launcher" in 08 and our "kicker" in 2010. Of course neither year required the precision that this year does. Judging from the videos... you've achieved it, though. Nicely done.

Hope I'm assigned to inspection on your field... I want to get a closer look!

Jason