Can you help us get our robot unstuck?

[Ben Wolsieffer]

We designed our robot to climb from one rung to another by attaching with an arm and then flipping over to attach with another arm. It works but only if we help it detach the lower arm after it flips. All the weight is on the lower arm and it can't let go of the rung to move on. This is a video of the robot climbing (but the human assistance in getting the arm detached isn't shown), http://www.youtube.com/watch?v=cgbJ7...ature=youtu.be. We can't figure out how to correct this problem. We thought of hinging the arm but that probably won't do it. Any ideas?

[Garten Haeska]

Not sure but it looks like you would be exceeding the 54in rule.

[Kevin Sevcik]

That's certainly an exciting way to climb. If you guys pull that off I think it means I lose a bet with a former student. In furtherance of this:

1. Are you guys sure you're making a valid climb? It looks like you come awfully close to touching the second level while you're still touching the floor.
2. Remember you'll be doing this from across the playing field. Lining up and making those motions may be challenging.

You've definitely discovered the downside of using gravity and leverage to grab something. How hard is it to release that bottom claw? Could you maybe attach a short stroke 2" bore pneumatic cylinder underneath the arm to shove against the pyramid diagonal and push the bottom claw away? A 2" bore has 180 lbs of push at 60psi. Or a 2.5" bore cylinder with 300 lbs of push. Bimba will sell you a 2.5" bore, 3" stroke cylinder for $90 each and ship it the same day.

[Kevin Sevcik]

Originally Posted by Garten Haeska

Not sure but it looks like you would be exceeding the 54in rule.

Once you're touching the pyramid, you're legal if you can fit in a 54" cylinder that's 84" tall, in any orientation. And you stay within 54" of the base of the pyramid. The latter is the the only thing they're in danger of. I think they can solve that by tilting the base vertical sooner, before that claw is fully engaged with the tower. It might require some frame adjustments so the wheels are still on the ground and they can still back away in a controlled fashion, though.

[engunneer]

There may just be a sweet spot where the tension on the lower grab comes off. Are you planning any position sensors on the arms so the program can move to preset positions?

Very impressive looking climber. I hope you work out the remaining issues.

[tsaksa]

From the video it looks like when you engage the second level the weight stays on the first level. This will probably always be true unless the distance between the two grippers is slightly less than the distance between the two bars. The grippers would then need slight tappers (or a lead in surface) to help transfer the weight to the upper gripper as it slid into position. I hope that makes sense. Basically a ramp to insure the weight ended up on the new grip point.

However any method of automatically transferring the load between grip points will be very tricky. Since the only holding force seems to be gravity and you are in the process of switching between hold points while your weight is cantilevered you have a significant risk of falling. I think you may need a solenoid driven locking pin on the grippers to prevent letting go two early. That coupled with a lead in and closer spaced grippers might do it. Or having another way to transfer the weight might work, such as some way to expand and contract the distance between the pivot points.

Whatever you do this is already a very unique way to climb. It will be great to see it in competition if you work the bugs out.

[DELurker]

It is an impressive robot. However, as other posters have asked, I'm wondering if it fits in the 54" cylinder with the bumpers mounted.

However, that's not what was asked. To detach the lower arm, your could grab the pyramid leg and give a little push. Alternatively, could you modify your gripper design to add a flange in the vertical direction so that it wraps behind the horizontal poles to take the load?

[Tem1514 Mentor]

I love the design of flipping end over end and will to try and suggest a possible answer that may or may not work is, using the video as a reference.

It may be possible to rotate the robot frame CW around the upper arm and at the same time rotate the lower arm CCW. This will have to done with great care so that as the load is changed from the lower to upper arm you robot doesn't just lose it's grip and fall off

Aside the 54 " possible problem you need to watch out for differences in height from one rung to the next which according to rules may be up to 1/4 inch but I would allow for at least 3 inches either way.

[tsaksa]

Originally Posted by Tem1514 Mentor

I love the design of flipping end over end and will to try and suggest a possible answer that may or may not work is, using the video as a reference.

It may be possible to rotate the robot frame CW around the upper arm and at the same time rotate the lower arm CCW. This will have to done with great care so that as the load is changed from the lower to upper arm you robot doesn't just lose it's grip and fall off

Aside the 54 " possible problem you need to watch out for differences in height from one rung to the next which according to rules may be up to 1/4 inch but I would allow for at least 3 inches either way.

Yes, I agree that this is what probably needs to be done, but if you study the contact points you will see that it may not currently be possible. Trying to rotate the robot frame CW around the upper arm will tend to push the robot frame down. And trying to rotate the lower arm CCW will tend to push the robot frame up. Both are not possible at the same time, and I think this is what ends up jamming the robot in place when it tries to change grips.

The design may need either an additional joint in one of the arms, or some ability to slide one of the pivot points a bit to allow this transfer to proceed smoothly. As it is, there may not be enough degrees of freedom in the system to allow the correct motion that can transfer the weight properly.

[thinker&planner]

It seems like there should be a "sweet spot" where it should release. You jut have to keep testing to find it, but you might consider opening up the grabber/hook part so that it is a little looser. This could be dangerous but might be enough to free it.
An alternative to the piston (If I were on your team, I would rather not add pneumatics for just one thing if you can do it with a motor) would be using a smaller motor (BAG or Bane Bots w/gearbox) and a cam/eccentric wheel to push away from the pyramid corner as you lift the bottom claw/gripper.

Regardless of how you end up doing it, you should definitely program the climb to be autonomous. If you find a sweet spot, it would be completely necessary to automise the climb.