Designing a climbing mechanism for 2013... a humbling experience

[pfreivald]

Originally Posted by Jon Stratis

In fact, the level of complexity to go from 2 to 3 is actually less than that required to go from 1 to 2 with this design!

That's true for us, too. Once we're at two, going to three should be essentially a repeat of what we just did... (And to answer the question, we should be able to climb an arbitrary number of levels given time and battery life.)

[Pendulum^-1]

We think we have a solution. Of course, I thought that a few days ago, also.

The challenge is significant, the answers will probably be subtle adjustments to a solid concept. The biggest problems are initial docking (we want it to be simple, quick, robust, and tolerant of poor angles of approach), transitioning from level 1 to level 2 (once you have that, 2-3 is rinse and repeat), and getting all the way up to the top of level 3, so that you are totally above level 2.

Rolling off the post and flipping over backwards are certainly problems. CG will be critical in the climb.

And you must really double check all the geometries out.

Is it risky? Yes.
Is the risk worth the reward? We think so.

[Craig Roys]

Another challenge for some may be the 54" radius. Unless I'm misreading from the Q&A, that is with respect to the floor - so if your robot inverts, it needs to do so inside of a 54" cylinder. They certainly didn't make this a trivial challenge this year.

[Andy A.]

Originally Posted by Craig Roys

Another challenge for some may be the 54" radius. Unless I'm misreading from the Q&A, that is with respect to the floor - so if your robot inverts, it needs to do so inside of a 54" cylinder. They certainly didn't make this a trivial challenge this year.

This was a question posed to the Q&A a couple of times even and, yes, the 54" cylinder is always with respect to the floor, not the robot. If your climbing scheme includes rotating the robot like the video posted earlier in this thread, you'll need to be mindful no part of your robot breaks the cylinder, including bumpers.

One of the more challenging 'size' restrictions we've had.

[falconmaster]

Here is our plan that we are trying to flesh out
https://docs.google.com/file/d/0B44TPv-TccPZTTVWLUV0Z0JlM1U/edit
Hope it works

[billbo911]

Our concept it somewhat of a "Monkey Bar climbing" approach.

Here is a quick video I threw together yesterday to demonstrate the basic concept to our team. Please excuse the extra enthusiasm I exhibit towards it's potential success, I'm just excited we have something that might work.

http://youtu.be/9kxMmYTVo0E

How would you rotate/tilt the bottom half of the bot without applying torque to the horizontal pipe. Tilting or rotating the bottom half of the bot on a unfixed upper point would just change the c.o.g and swing it under the bar further. I think ???

[Donut]

After pounding our head at it a few days, we have 2 designs that we think are within our team's fabrication abilities. I would argue that only one of them is simple but that is an opinion. We're building prototypes of them now and won't know how truly well they work until probably Tuesday next week.

I think "simple" is a relative term for the 30 point climb. A "simple" mechanism for climbing could still have quite a bit of movement.

[cmrnpizzo14]

Originally Posted by danopia

For those of you that have ideas for a 30-point climb, would your robot be able to scale to a pyramid of, say, 10 levels without nontrivial changes? Ignore your robot hanging 30 feet in the air.

Yes, we have a design that is simple, fits a small footprint and potentially will climb to the top in <20 seconds. It should in theory be able to climb to any level without any change.

[ttldomination]

Originally Posted by cmrnpizzo14

Yes, we have a design that is simple, fits a small footprint and potentially will climb to the top in <20 seconds. It should in theory be able to climb to any level without any change.

Funny thing about that theory...

- Sunny G.

[cmrnpizzo14]

Originally Posted by ttldomination

Funny thing about that theory...

- Sunny G.

If it can make it to the top when we build it, it can make it to any level. The design does not require changes based on levels, we just have to be careful when building it. A simple error when building could provide quite a headache for us when trying to climb.

[Jon Stratis]

Originally Posted by danopia

For those of you that have ideas for a 30-point climb, would your robot be able to scale to a pyramid of, say, 10 levels without nontrivial changes? Ignore your robot hanging 30 feet in the air.

It seems like some postings I've seen would because it's repetitive, and others have like a full dance routine going on that lands perfectly on top and wouldn't work for any other pyramid.

On Friday, we had a design that could do 3 levels and would need changes to do more. I leave for the weekend, and now the team tells me we have a design that could do 100 levels if we wanted, without any changes

[Madison]

I wonder how many of the 'simple' ideas people profess to have were things we considered and dismissed as being too unreliable, too unpredictable or not simple at all.

We have shied away from anything that relies on the machine's center of mass being in a known, good location or from manipulating that center of mass to encourage the robot to swing into some other orientation. There are too many unknowns there for me to be comfortable putting considerable effort into such a system.

We have some ideas for climbing up the corner that do not require manipulating our CoM, but they have their own challenges. We are likely going to focus instead on frisbee manipulation -- the potential for incremental improvement in that aspect of the game seems more promising to us than the potential, fixed 30/50 pt. contribution.

[Cory]

Originally Posted by Madison

I wonder how many of the 'simple' ideas people profess to have were things we considered and dismissed as being too unreliable, too unpredictable or not simple at all.

Anyone who thinks a 30 point hang is simple has no idea what they're talking about, IMO.

There are so many dependencies that affect every aspect of your robot. This is the hardest task we've ever been asked to do and I will bet money there are no more than 30 teams in all of FIRST that do it in an official match.

[connor.worley]

Originally Posted by Cory

I will bet money there are no more than 30 teams in all of FIRST that do it in an official match.

And far less will do it in a reasonable amount of time.