Putting robotics at your service™

Free shipping on orders over $200

Kurt's conversion of CHR-3 to TH4-R with 2.1 beta source

Print view Share :
Previous topicNext topic

Page 1 of 3 [ 34 posts ]

1, 2, 3
User avatar
Guru ( offline )
Posts: 4903
Posted: 2011-10-04 17:40 
As I mentioned in previous post, I am in the process of converting my CHR-3 into a HR-4, with T-Hex 4DOF legs. So far I have converted 2 legs. I also hacked up a beginning of a CFG file for it. Still needs lots of work. But for laughs...



Also I have found a few bugs that I have fixed in the 4DOF code, will do more testing once hardware is working better!

Edit: Not sure if I will have any updated videos of the part 3 and part 4 DOF robot as part of one leg fell apart and there is one servo in one the legs that I have not rebuilt that has a servo with bad gears... I think I have 2 or 3 spare 645s to fix it...

Since the video, I have updated the 4DOF code to handle the 3DOF legs exactly like when 4DOF is not defined.

Kurt

Edit: Changed Name to TH4-R to better match the naming convention :D


User avatar
Guru ( offline )
Posts: 1130
Posted: 2011-10-05 02:47 
Looks like he needs to get a little bit more familiar with his new shoes ;)

_________________
[b]Share, Use and Improve![/b]
Digging trough: [i]Theory of Applied Robotics: Kinematics, Dynamics, and Control[/i]
Xan's YouTube Channel
http://www.youtube.com/xanore


User avatar
Guru ( offline )
Posts: 2120
Posted: 2011-10-05 04:28 
Looking good Kurt. :D
I do see that one of legs are more or less "broken" but he walks though. :wink:

Interesting to see a 3/4 DOF combo in action. Also a combo of different length of leg-segments needs to be tried one time. That could be cool on an even more nature-like ant robot one time. With very large rear legs and so on.

A little tip is to place the camera a little closer so that we can see the legs better.

Thanks for posting the video! :D

_________________
Kåre Halvorsen, Zenta
-----------------------------------------
Zenta's blog
http://zentasrobots.com/
Zenta's YouTube channel
http://www.youtube.com/ZentaOlbaid
-----------------------------------------


User avatar
Guru ( offline )
Posts: 4903
Posted: 2011-10-05 09:25 
Thanks Guys :lol:

I was going to zoom in, but then first attempt the battery ran out of power :oops: and then after replacing the battery, the servo was really hanging up and then the leg part fell off...

I am in the process of now updating the middle legs (if there is such a thing with a round Hex) to be 4DOF, which replaces the bad servo and the leg parts that fell off. I have the two legs mostly assembled now, just finishing building a servo extension wire, put in the Horizontal Hip servo and mount them, and then doing the zeroing...

The bad servo feels like it is missing a tooth or two in the gears...

Now back to my coffee.

Kurt


User avatar
Guru ( offline )
Posts: 4903
Posted: 2011-10-05 11:09 
To follow up on:
zenta wrote:
Interesting to see a 3/4 DOF combo in action. Also a combo of different length of leg-segments needs to be tried one


Yes, that was my thoughts as well, which is why I started to do some testing of the combinations. My first idea was to change the T-Hex to have the 4DOF on the front legs, but I want to keep the T-Hex in the default configuration for now, so we can test TA code.

But something like the A-POD with size legs might be interesting. Like maybe the middle legs be shorter and 3dof but the outside legs be longer.

Another interesting complication is the Y zero point is different between the different leg designs, so it is sort-of trial and error to get the right Initial positions. Probably can simply use Caliper to measure delta...

I was also thinking about trying the CHR-3 with 4 4DOF legs and 2 CHR-3 type legs, but not sure if worth time to rebuild the 2 legs into that configuration, when my end goal is for all 6 to be T-Hex 4DOF legs...

Anyway lots of fun.

Kurt


User avatar
Guru ( offline )
Posts: 4903
Posted: 2011-10-06 13:58 
Update, I have found some other issues with the code and the T-HEX 4 legs. I have made changes to handle it :). I now have 4 out 6 legs converted to 4DOF.

Here is another poor quality video showing my THR-3/4 (Not sure what to call it now as it has different legs than any of the .HR bots :lol:



I will update the code in the beta soon. Also let me know if anyone is out there with a T-HEX with 4DOF that would like to help test out the 4DOF code.

Kurt


User avatar
Guru ( offline )
Posts: 3298
Posted: 2011-10-06 14:33 
yes you can see there are contact problems with the 3dof legs when rotating and also a little when walking. its looking a big brute now with those 4dof. i like the THR-3/4 mix.

making good progress. 8)

_________________
Jonny Poole,
Freelance Concept Art & Animatronics, 3D Art and 3D Modelling, as well as Physical Modelling, Mechanical design, and illustration.
http://www.innerbreed.co.uk
Create, Innovate, Inspire.


User avatar
Guru ( offline )
Posts: 4903
Posted: 2011-10-06 22:49 
Thanks,

Yes, I have not taken much time to try to resolve some of the configuration issues of the 3DOF legs, as they are non-standard and they will soon be replaced with 2 more 4DOF legs. But I thought I would at least test test the idea of mixed length legs. If I were going to keep the 3DOF legs, I would probably rebuild them and replace the short C brackets with Long C brackets, like what are used in the 4DOF legs. I would then move the mounting of the C bracket down to the lower hole in the GP Bracket, then both sets of legs will have the same starting Y location. Soon to convert. Although I am trying to decide if it would be fun to try the old CHR-3 legs and really look different. Especially if I don't find the other 1.5" tube that should be floating around here.

Kurt


User avatar
Guru ( offline )
Posts: 4903
Posted: 2011-10-09 12:51 
I split this off from the Beta 2.1 code thread as a lot of this is my play time :D and thought it might be better to leave that thread for posting new code changes and the like. i.e. when I am happy with the 4DOF code, I will put up new zip files with code to support it.

Last night I finished the conversion of the xHR-3 hex robot to a THR-4 hex. That is all 6 legs now have 4DOF. I have the code setup now to use a new concept of ServoHornOffset (introduced by Zenta in his 4DOF T-Hex (when not using software offsets...), to allow us to the SSC-32 offsets which are limited in how large they can go...

I followed the tutorial (http://www.lynxmotion.com/images/html/build168.htm) to build the legs and then used the 4DOF T-Hex tutorial to do the servo offset zero procedure (http://www.lynxmotion.com/images/html/build171.htm)

I am wondering if my legs are correct. That is if I run my updated software, the initial leg position (when hit start button on PS2) looks like:
Attachment:
THR4-Leg-Initial-position.jpg
THR4-Leg-Initial-position.jpg [ 28.18 KiB | Viewed 1621 times ]


If I measure how far out the leg is at this point, I have:
Attachment:
THR4-Leg-Measure.jpg
THR4-Leg-Measure.jpg [ 15.67 KiB | Viewed 1621 times ]


Note: I played around some with the initial positions which are currently set for XZ=70 and Y=50. To see how this compared with using the T-Hex 4 software, I simply ran that software on the HEX. Yes I expected that the servos would be off some as the angles of where the legs are is different. Here is a picture of how that looks.
Attachment:
THR-4-with-THex4-Software.jpg
THR-4-with-THex4-Software.jpg [ 18.71 KiB | Viewed 1621 times ]


Does this look right? Does someone have a THex-4 that they can take a picture of when the robot is in it's initial position?

Thanks
Kurt


User avatar
Guru ( offline )
Posts: 3298
Posted: 2011-10-09 13:42 
hi kurte,
congrats on the conversion. looking forward the the video. :wink:

when i was setting up my 4x4 i used the setup zenta used.
viewtopic.php?f=8&t=6521&p=64839#p64839

init pos:

_________________
Jonny Poole,
Freelance Concept Art & Animatronics, 3D Art and 3D Modelling, as well as Physical Modelling, Mechanical design, and illustration.
http://www.innerbreed.co.uk
Create, Innovate, Inspire.


User avatar
Guru ( offline )
Posts: 2120
Posted: 2011-10-09 15:06 
Hi Kurt,

Your circular T-Hex version looks great! I see that you also have "the book".. :wink:

A little tip, compare the IK angles in the code with the actual angle on the legs. If they are very off you should get a hint of whats wrong/off.

_________________
Kåre Halvorsen, Zenta
-----------------------------------------
Zenta's blog
http://zentasrobots.com/
Zenta's YouTube channel
http://www.youtube.com/ZentaOlbaid
-----------------------------------------


Expert ( offline )
Posts: 264
Posted: 2011-10-09 15:46 
Hi, Kurt,

I have been following your conversion with interest. Now that you have 4DOF on all six legs it may be too late to ask my questions, but I am wondering:

In both of your videos, 4x3DOF+2x4DOF and 4x4DOF+2X3DOF, the odd pair of legs are adjacent. Did you ever try it with the odd pair diametrically opposite each other?

If you considered it but rejected it, was your rejection because you anticipated severe programming difficulties, or because it would delay your goal of the full 6x4DOF conversion?

Looking forward to more videos.

Thanks,

_________________
Ted, RoboTed
Never quite finishing means never having to admit failure!


User avatar
Guru ( offline )
Posts: 4903
Posted: 2011-10-09 15:53 
Thanks, Yep I have the book. The only problem is that in the 30+ years since my last college math class I forgot everything, so it all looks Greek to me...

Looking a bit more at why I think my legs do look right to me, may be an issue with the THex 4DOF manual. This is if you look at:
Quote:
Step 12.
Also adjust each leg's knee outward one 'click' or 15°.


But if you look in the THex-4 source code at the offsets, it says:
Quote:
cTarsHornOffset con 150 ;Snap out the horn one click inward

So my guess is Step 12 is wrong and my leg is off by 30 degrees.

Will try to update it soon!

Once I have it feeling right, yep I will create another poor quality video :lol:

Ted: My first thought was to have the other legs be the two middle legs and it would be interesting to do. I only did it the way I did was because while making the first video I found that there was at least one bad servo in the middle legs, so I rebuilt those next.

Kurt


User avatar
Guru ( offline )
Posts: 3298
Posted: 2011-10-09 17:13 
kurte wrote:
Quote:
Step 12.
Also adjust each leg's knee outward one 'click' or 15°.



But if you look in the THex-4 source code at the offsets, it says:
Quote:
cTarsHornOffset con 150 ;Snap out the horn one click inward


Kurt


Step 12 is for the Knee outward one 'click' or 15°
Code refers to the Tar inward one 'click' or 15°

both joints require the 15º +- click

Zenta pointed out the easiest way to solve the calibration part is to first set all servos to the center 1500 position and mount the horn so that the leg pose the 0-deg positions (like picture #1 of the link i posted). Then snap out the horn from the femur (Knee) and turn it 15 deg upward. Then snap out the tars horn and turn it 15 deg inward. Hopefully the leg should look close to the second picture.

_________________
Jonny Poole,
Freelance Concept Art & Animatronics, 3D Art and 3D Modelling, as well as Physical Modelling, Mechanical design, and illustration.
http://www.innerbreed.co.uk
Create, Innovate, Inspire.


User avatar
Guru ( offline )
Posts: 4903
Posted: 2011-10-09 17:55 
Thanks Jonny,

Yep, I should have re-read over Zenta's thread, but thought it would be good to follow the tutorial... Where Steps 11 and 12 with figures 11 and 12 show moving both servos joints out 1 click, which ends up making the foot at about 30 degree angle.
Quote:
Step 11.
Using the same method, adjust each leg's first vertical servo upward one 'click', or 15°. The long 'C' bracket should now be parallel to the ground.

Step 12.
Also adjust each leg's knee outward one 'click' or 15



Actually looking at the picture again, I think it shows it moving it inward, but the text says outward and it is also is not the knee...

Zenta and your description makes more sense. It is also why I wanted a picture, showing the robot at the init location to confirm... Thanks Again...


1, 2, 3

All times are UTC - 5 hours [ DST ]. It is currently 2014-08-29 10:18
Feedback Form
Feedback Form