Bioloid cartwheel?

[Apollo]

I am considering Bioloid for my first biped robot. I have seen cool videos of RN1 and Kondo robots doing cartwheels and was wondering if the Bioloid could do this? Are the servos strong enough?

Ross

[Joe]

I am considering Bioloid for my first biped robot. I have seen cool videos of RN1 and Kondo robots doing cartwheels and was wondering if the Bioloid could do this? Are the servos strong enough?

They certainly should be -- they're about 30% stronger than the ones the Kondo robots use, which themselves are stronger than those in the RN1. So it's probably just a matter of working out the motion sequence.

Best,
– Joe

[Apollo]

That is what I suspected. I would love to hear from a Bioloid user who has attempted these types of movements. I think the video clips showing Kondo and RN kits doing cartwheels and other acrobatic exercises went along way for me in the appeal of a Biped robot. It would be cool to find more Biloid videos on the web. I am still leaning heavily towards Biloid as my first kit.

Thanks for your reply.

Ross

[jaho434]

Ok, the challenge has been thrown down.

First person to post a video of their Bioloid doing a cartwheel gets...... er, kudos and bragging rights.

I'll give it a go tonight.

[tempusmaster]

Ok, the challenge has been thrown down.

First person to post a video of their Bioloid doing a cartwheel gets...... er, kudos and bragging rights.

I'll give it a go tonight.

Okay - That would be great to see. I'll sweeten the pot. The first person that posts a video of their Bioloid doing a cartwheel on Youtube or Google Video gets it featured in a post on Robots Dreams.

[jaho434]

I'm working on it! It's a little harder than I first imagined.

[beermat]

Don't know if you are referring to hiveminds RN move, or if hivemind lurks on here, but if you are referring to hiveminds move, then I had a go at recreating this for my RN after his video was posted but before his code was posted, so I got to learn some of the issues involved in creating this move.

The most important thing we both found with the move to make it work (and make it repeatable) was a key timing issue. Basically, if you watch the video you will notice the RN folds down quickly at the waist and raises its feet. This has the effect of tipping him forward onto his hands. I then continued to program the rest of the move to flip his legs over right after, and could not get it to work at all. In experimenting with the first part of the move - landing on the hands - I left him in that position and noticed that if I nudged him from behind gently, he would quite easily rock forward and rotate over his hands and basically have his weight and momentum going in the right direction to finish the move. The movement onto his hands actually creates enough forward momentum for the tip over, but the key point is, it isnt instantaneous after the completion of the move code. Once the servos are in position, there is still a bit of time required for him to land and rock over his hands before he can even start the next part of the move. A simple delay after the first part and before the second part was enough to make this move work. 150-250ms delay seems to be the optimal value for a RN.

So, for the RN, this wasn't quite as simple as a structured set of move here, move here, move here type code, some allowance had to be built into the routine to pause and let momentum do its thang. Hope this helps!

In addition, I actually managed to use almost exactly the same code in the RN to achieve something completely different. See the results here

[Apollo]

This is a very cool move! The move we had in mind was a cartwheel which is where the biped falls sideways onto it's hands with enough momentum to 'roll' back onto it's feet. I imagine this is a pre-programmed move for RN. My understanding is that the Bioloid kit does not come with this move pre-developed so the challenge here is to make it work and post a video.

I have just ordered my Bioloid kit so I am looking forward to attempting this myself!

Cheers,
Ross

[beermat]

For some reason, I was thinking of the forward roll when I read cartwheel. Do you guys have access to the RN code for the RN cartwheel, and would it help? I can post it here. It is pretty cool.

[hivemind]

Yea, actually I tweaked the RN(stock) cartwheel move a long time ago so that it would be faster and more reliable, I can tell you how I accomplished that if you would like to hear.

I suppose my first words would be that you dont want to be cautious about the speed. The momentum and delay in the middle are key. In the example that I explain I will be rolling to the left (so I will mentioned left and right arms) hopefully it will not be too confusing.

So first I had the robot sink down quickly and raise both arms up to shoulder level. The robot should be half way between squatting and standing.

Next he would lean to the right at about 45 degrees and extend his right arm. The left arm would raise above his head at that point.

Quickly he would shift all weight to the right, and his right arm will come back closer to his head. (at this point he should be falling to his right) The left arm should now go down until it almost touches the ground-but be sure to make the arm fully for balance and guidance and not to slow down the motion.

After he gets here the momentum is in full swing working for you. Delay all movements for about 200 ms except for letting the arms get a nice curve to them.

The legs go to a squatting/sitting position and the right arm now pushes off, while the left arm will do something similar(ie: push off somehow depending on the position). If the momentum is already helping you past here then just adjust the legs to catch the robot on the other side.

Your robot should land on his feet, and a final (slow) flick from the hand (of your robot) should get him balanced and in a sitting (or squatting depending on the speed) position. I recommend a stand and bow.

Anyway, that is how I reconstructed it. The tough part for me was making sure I had the hands in the exact right position. Even a couple degrees make a huge difference here. But once you have the code right it works every time even on different surfaces. Surprisingly it is very reliable.

[Joe]

C'mon guys... Nobody can get their Bioloid humanoid to do a cartwheel?!?

Cheers,
— Joe

[jaho434]

Been busy this week, but have a few days off, so will get back to it!

[Joe]

Still no cartwheels? I begin to worry that maybe the Bioloid kit can't do it. The servos are supposed to be quite strong, but maybe the brackets are too heavy?

Or do you think it's just a tricky motion sequence, and it's just a matter of working it out?

Thanks,
— Joe
(still sitting on the fence about what kit to get)

[jaho434]

Well, I still think it's possible.

I've only had a quick go at it and got it to the upside-down point, but then it toppled over. It's just a fine line with the balancing to keep it upright while it's rolling.

[Joe]

Just checking in, over a year later... if nobody's managed to make their Bioloid do a cartwheel by now, I think we can safely say that it's probably not possible.

I'm mighty curious as to why that would be, though. Can anyone post their experiences in attempting this?

Thanks,
— Joe