Double Catapult concept robot for Clean Sweep

Link to pictures forum:
This robot was at the STEM college table at the 2009 wold championship for most of two days. I gave a couple people guided tours of it, but decided to post it here also, as pay it forward for RFolea’s catapult pictures.

This is iteration zero of an unfinished concept robot for Clean Sweep. The initial catapult concept generated from pictures posted by RFolea. Observation of original catapult video shows fire cycle time is 6 seconds. With double catapult concept, fire cycle can be reduced to 3 seconds. Half rotation time is less than 3 seconds. Holonomic drive rotates particularly efficiently.

  • Steel backbone, then all aluminum, 8lbs without battery.
    Could have been nearly all aluminum (except misc gussets and linear slides) Backbone is half-lapped into frontback
  • Low CG: 9 motors, all within 1" of the floor, most do not need extension wires to get to controller.
  • Small omni-wheels (following lead of Exothermic Rick Tyler) with 36:12 gearing; doesn’t work that well, haven’t driven it since upgrading to double rollers.
  • Omni-wheels shaved width, double roller mod, needs better craftsmanship to get rollers offset from each other as done in 4" omniwheels.
  • Holonomic X drive: lots and lots of 45 degree angle gussets!
  • Inverted rack&pinion& pin catapult latch release using linear slide, should use potentiometer for feedback. Currently releases only one side, could do both with some work to jag around congested winder motor sprockets.
  • Hinged catapult arms, using a hinge cut in half; needs anti-pivot reinforcement plate on arm side of hinge.
  • Hinged scoops fold up to start in 18" cube, either with arms flat, or more likely, both sets of arms at 45 degree angles.
  • Lightweight scoops with zipties, needs more work.
  • One side indented to fit around triangle goal for some misc future concept not yet developed.
  • Central tower powers catapult springs on both sides.
  • Four motors in base on circular high strength chain fix some of the short-coming of original catapults.
  • Off-center mounted second HS chain has rubber springs at each end.
  • Goal of sprocket eccentricity is to provide more constant torque to motors; When spring is lax the large radius of gear winds it more quickly; When spring is tight, the small radius winds it slowly, eases motor strain.
  • Small chain ratioed potentiometer would provide limit feedback to spring winder system. Needs limit switches under catapult arms to show when they are down enough to engage trigger pin.
  • Vertical central controller position allows good access to battery/cables/switch, without getting in way of double nested catapult arms.

Intended strategy for this robot:

  • Lose autonomous, (on purpose if necessary), this gives your side the control of most of the balls.
  • Spend most of match looking helpless while pushing balls into neat rows against back wall. If competing in college, make sure not to lose early via “clean sweep rule”
  • At 30 second bell, start playing in earnest. Scoop up against back wall, fire, rotate, repeat every 3 seconds, for 10 repetitions. Possibly use autonomous recorded sub-routines to assist.
  • With one robot, 8 repetitions x 18" are needed to clear 144" of back wall, so there is some small margin for imperfection. Second robot can help stack balls to rear wall, or a college twin can help clear the field at 20 seconds left, rather than 30 seconds left.
  • Goal is to launch balls all over the opposing field faster than opponents can chase them down to pick them up and redeliver.
  • This strategy would have led to some exciting finishes: a blizzard of flying balls vs the strategic last mega-dump.
  • May have required a nested strategy, using hold-back strategy as outlined above only in finals.

??If you were an opposing dumper team, what would your strategy be?

??Would you really change your autonomous to grab control of the balls rather than pushing them over?

My strategy would be exactly the same as if I were facing any other robot. I think you severely mis understand the abilities of dumpers. You strategy sounds good, but I’ve seen excellent catapults that have simply been overwhelmed by the onslaught of balls that dumpers can give.

Not to mention, with your locking system, I believe that most of the strain is being put on the motors, and you’re going to burn them out before you know it.

Motor burnout would be something to consider during iterations to find the best tradeoff between winder speed and spring power. The current setup is 4 motors with 3:1 at peak winding speed, and 10:1 at the holding point. I am unable to move the chain by hand with the power off. With the final strategy, the winder motors are only used in the final 30 seconds, and only at the holding position for 10 seconds.
I don’t believe I am underestimating ball delivery of dumpers, (other than the high goals, which I didn’t have a counter plan for). With this strategy, the dumpers have nothing to do after they nearly sweep the field in the first 30seconds. I may be underestimating the pickup speed of the good dumpers though.
I didn’t notice excessive hold-back strategy in Dallas. There were many matches where the side that was behind would start dumping out the back. Then usually the side that was winning would NOT finish clearing their side. At that point, either side was capable of winning by having a more successful last second dump.

but if they see what you are doing
they may keep a bucket of 6-7 balls also for the last 30 secs

I’m getting a more careful look at your catapult bot and have a few questions. 1) Where do your rubber bands attach?
2) I see how that arm flips backward to create tension, but how does it release?
3) I noticed a sensor in the center of the 5-wide c-channel (potentiometer I assume) and was wondering how you used it. Perhaps to determine when the arm has reached the position for full tension?

Hope you don’t mind giving away a few (more) of your trade secrets.


  1. There is a free piece of HS chain on the eccentric sprocket on the top, with rubber bands at each end connecting to the catapult buckets. There is a loop chain that is driven by multiple motors at ground level, to rotate the top sprocket, which is bolted to the eccentric sprocket that engages the free chain. The point of the eccentric sprocket is to provide large displacement at low force at the beginning of the tensioning, and low displacement at high force and low holding torque requirement at full tension.
  2. The left arm is lowered by running the free chain in that direction. If the right arm is not already latched by a linear-slide-pin, the right arm will raise up at this time. When the left are is fully down, the linear-slide-pin can engage the bucket to hold it down. Now the loop chain can drive the free chain to stretch=tension the left arm, while lowering the right arm. When the right arm is fully down, the left arm is fully tensioned.
    This is iteration 0.5, the arms are only lowered by gravity.
    I have only installed the linear-slide-pin on one end so far; there should be a view that shows it. It needs an extension to reach the other pin to hold the other bucket, but the central tower area is congested with the HS loop chain and sprockets, so the extension would need a C-offset. The same linear-slide-pin motor should be able to operate both arms, since only one side needs to be latched at a time.
  3. The sensor (pot) is on a reduction chain sprocket because the top axle rotates more than the pot will allow. The point of this sensor is to know the location of the free chain, so that it does not get rolled all the way off either end, which is also the point of full tension, as you said.

The “double ended” concept might be useful for Round-up.