Felix, a SES based 4x4 DOF quad

Hi!

As some of you know I have been working with a new robot project. This time I’m trying to make a quad walk. I must say when I saw the Big Dog on youtube I was VERY impressed and thinking that must be almost impossible to copy :laughing: Anyway I’m not trying to say that my goal is to make a “mini Dog” or a quad that has the abilities like this big dog (its probably way way over my little head :unamused: )but some of the design are inspired of the Big Dog.

One big challenge with all quads is the walking part. Its almost like having a four legged table, cut of one leg and its very unstable. And the table has not even tried to move :laughing: . So I have been thinking of a leg design where you easily can shift the COG, by adding one DOF. Another challenge with some quads (2 DOF) is the turning part.

Here is a picture of the body (this is an earlier version, a bit shorter) http://forums.trossenrobotics.com/gallery/files/1/5/3/5/felix_body_v1_001.jpg
As you can see I’m using a 5645 servo and this is the 4. DOF. The body are made of 3mm aluminium, all parts are handmade with a bandsaw.

For powering I’m using 2x LiPo’s and I’m trying to use 2 Turnigy 8Amps (maks 15 amp burst) DC-DC converter. This is a picture showing them modified with a spacer I used for mounting:

Another picture of the brackets/spacers for holding the servowires and the powerswitch:

And here they are mounted to the body. I’m gonna place the SSC32 on top of the black plastic plate.

Another view (underside):

This is a picture (also bottom view) showing the Turnigy mounted between the servos:

And finaly a complete mounted body with the SSC32 onboard. I still have some work left. Calibrating, mounting batteries and BlueSMIRF.

Another view:

Legs curled up:

For the time I’m also working with the kinematics on a Excel sheet. The reason for why I’m using Excel is that its very easy to experiment with different fomulas and kinematics methods. But my goal is also here (like for the Phoenix) to have a onboard brain doing all the math.

Felix is not walking yet…

BTW: The foot are made of parts from my very first hexapod I made late 2006:

I’ll post more information when I have a video ready :wink:

Wowzers. That thing is very professional looking. It’s awesome how you got the top DOF tucked up inside the body.

You make cuts like that in aluminum with a bandsaw? You have mad skillz!

wow. think this might be the first thing people say or think when they see the work you’ve done.
looks really nice, well done.

i idea of using the extra dof really works well. i can see it walking already.
i also have been wirking on a quad. but with 3dof per leg. tricky! lynxmotion.net/viewtopic.php … sc&start=0

cant wait to see the vid?!

zenta, Felix is a beautifully crafted machine! Can’t wait til you apply your IK to this thing… I guess more then IK, you’re gonna have to have a sequencing scheme between the moves right? The sequence of when a four-legged animal trots and when it dashes/runs is quite different. I don’t think one sequence will work for all cases for a quad, that’s what I find to be the most challenging thing about quads…

For hexys, you can theoretically speed up your gaiting and just pump out the sequences (aided by the IK calculations) faster and you’ll have a faster walk, but not so with the quads!

Can’t wait for the videos! :wink:

I like the ‘nesting’ of those two servos at the hip! Very clever!

Of course the entire bot looks good, but we wouldn’t want to inflate your ego too much would we by saying it?

(Kidding! I want to inflate your ego even more so you make more cool looking bots!!)

Hi KÃ¥re,

Very nice indeed! No wonder you wanted to wait to release the first pix!

I’m very impressed. Almost makes me want to abandon the the ideas I’d had of late of my new 'quad. I’ve only been considering a 3DOF leg. And a considerably lighter construction. One that could be DIY by most anyone.

I’m going to need to continue thinking about my initial 'quad design.

Alan KM6VV

Zenta,

One word: Awesome!

I saw some pictures before on trossen. You really know how to “driveâ€

You can put me on the very impressed list!

I really get a kick out of seeing people make things from my products that I never imagined were even possible!

Very nice quad KÃ¥re! 8)

Zenta - looking good so far! I’ve been thinking of the COG shift issue with quadrapods. What if you had a continuous rotation servo that was responsible for shifting a counterbalancing weight equal to a leg and then had this servo rotate so that the weight would always be opposite of the leg that is lifted?

Thanks to you all for your kind words, it still means a lot for me!
Hmm :confused: still my wife are not especially impressed, just an "ehm, okay. What can it do? :unamused: " :laughing:

Yes, your Stalker is impressive! I was hoping the 4. DOF would help doing the COG shifting easier when standing on three legs.

I’m still working with the IK part. I’ve solved parts of it. But the walking part are much more demanding, I can’t use the same straitforward sequencing like for an hexapod because of the continous COG shifting.

I’m taking one problem at the time. First I have to finish the building part, then calibrating and the SEQ setup, then I need to really dig into the kinematics and walking part. I’m still a bit unsure if I can make it…

Thanks Alan! Don’t abandon your idea, stick to it! I’m sure you’re gonna figure it out. A lighter construction sounds reasonable. My Felix is a bit heavy…

I’m not going to make Felix jump, the servos are not fast or strong enough. Have been thinking of the jumping part, but I’ve not figured out how. I tried to make Phoenix jump by adding some really strong springs to the femur, but it was not a success :laughing:
When or if I come to the point where I’m gonna add an onboard brain I’ve to add some sensors I think :wink:

Similar idea has crossed my mind too, but I found it a bit difficult to make space for it and to be honest I don’t think a counterbalancing weight are equally efficiently than using the whole body for balancing.

Hi,

I’m not sure what LiPo’s to use, I have these 2S 1700 mAh (total 3400 mAh):

And these large 2S 4000 mAh LiPo’s (total 8000 mAh :smiling_imp: and exact fit ! 8) ):

Of course the large batteries add weight, total weight for Felix with these batteries is little over 2,5 kg. And with the small ones about 2,2 kg. A little overview picture of them all:

I’m not sure if Felix is able to walk very well with the large ones, but I’ve got to try… :wink:

What do you think?

I say go with the large, they fit well and they’ll provide the energy needed for the servos to move the bot with them. 0.3 kilograms isn’t that big of a sacrifice to increase the mAh by over double of the smaller ones…

Coolest quad ever. Zenta needs to join the lynxmotion team :wink:. I smell a hiring in the near future.

Just use 1 of the larger packs, should be more than enough capacity and amperes. Do you plan to use a BEC/regulator or are the servos you have chosen ok to run on 8.4V (fully charged 2S lipo can reach 8.4V)?

I second that.

Thanks for your tip! I’ll think I probably have to experiment for what combination that gives the best walking result.

Using only one large pack was a good idea! I just need to make a Y-split for the two UBEC’s (if you look closer to the pics above you’ll see them Eddie :wink: )

Thanks!

Hi KÃ¥re,

Oh, I might delay the project, but I’m not giving up. It will work, but is it what I want? I had even been thinking of a 2DOF ‘quad. Cheap! But it would lack the elegance of 3 or 4 DOF legs. What all are you going to do with the extra joint? The sagittal “hipâ€

Hi Alan,
Sorry for the late reply.

Sounds interesting, some inspiration from G-dog? Or a quad version of Loki :wink: A 2 DOF would make it alot easier to give an onboard brain without to much kinematics and calculations.

Hi KÃ¥re,

G-Dog a little, but a little from the “French Dogâ€

Why dont you guys look at some of the videos for LittleDog (rather than BigDog) and try to program something close to him? Boston Dynamics programmed BigDog after him, I guess.