T-Hex, a Mech inspired 6x4 DOF hexapod

Lynxmotion Legged Robots
1

Hi,

Lately I’ve been working with a new hexapod project. This time I wanted to go a different way when it comes to design. I must say that all the talk and focus on the Mech warfare has inspired me too, therefore I choosed to try out a mech inspired body design. The legs are made of standard SES brackets and the only custom made parts are the two body panels and the spacers.

Part list (I’ll add the SES parts and more details later):
24 x 645 MG Hitec servos for all legs
2 x 85 MG Hitec servos for head (optional
SSC32 V2 Servocontroller
BB2 Botboard for BAP
BAP28
2x Turnigy LiPo 6v regulator
1x 5400 mAh LiPo
DIY RC remote (future plan to use XBEE)
RC receiver

The major reason for starting this project was that I wanted to make a hexapod with 4 DOF legs. The joint are called coxa, femur, tibia and tars. You may ask why I choosed to add an extra joint (the tars)? First of all this is more like an experiment and a proof of concept. A 3. link to the 2 link IK chain (femur + tibia) gives us an endless amount of IK solutions. So like a robot arm the tarsus act as an end-effector. The IK calculations are relative easy if the tarsus are defined to stay in a perpendicular position in relative to the ground. I’m using a variable called TarsToGroundAngle and as a reference I’ve set the perpendicular position to be TarsToGroundAngle = 0 deg (not 90 deg). So, if we wanted the tars to hold a constant perpendicular position to the ground the formula for finding the tarsus angle would simply be:

Tars = Femur - Tibia, or Tars = TarsToGroundAngle + Femur - Tibia

(The femur are calibrated with 0 deg in horisontal position and the tibia and tars are calibrated in a vertical position). This actual pose of the leg are illustrated here (screendump from PEP):

But using a constant TarsToGroundAngle wouldn’t always be the best solution for a leg. I therefore made an algorithm/formula for calculating the variable TarsToGroundAngle. The input values are the feet positions (coordinates). Here are some pictures that demonstrates how the algorithm work: (pay attention to how the tarsus varies in relative to the ground).

I’ll be the first to admit that adding the extra joint (tars) can be a vaste of weight, energy and cost. And its also a bit overkill when it comes to the IK part. But there are some advantages… The leg are way more flexible and are able to walk over large obstacles without the need for extending the femur. Here is a picture of the SES based leg:

The maximum ground clearance is about 18 - 19 cm (about 7 - 7,4 inches). My idea is to use T-Hex for experimenting on terrain adaption. At this moment I’ve not added any feet sensors, future plans though.

Here are some pictures of the main body:

I’m using only one battery (LiPo) for both electronics and servos. For those of you who are interested, here is a little tip of how a made the battery holder for T-Hex:
I used the little spring that are inside the Hitech plastic boxes:
forums.trossenrobotics.com/gallery/files/1/5/3/5/pb111420_thumb.jpg
Then I cut off the upper half of the spring clamp like this:
forums.trossenrobotics.com/gallery/files/1/5/3/5/pb111421_thumb.jpg
I made a custom spacer with a hole for holding the spring:
forums.trossenrobotics.com/gallery/files/1/5/3/5/pb111429_thumb.jpg
The spacer has a hole on each end, here are the spring mounted on the spacer:
forums.trossenrobotics.com/gallery/files/1/5/3/5/pb111432_thumb.jpg
And the spacers are mounted inside the body:
forums.trossenrobotics.com/gallery/files/1/5/3/5/pb111434_thumb.jpg
Battery holder with the battery in place:
forums.trossenrobotics.com/gallery/files/1/5/3/5/pb111439_thumb.jpg
Another view of how the spring holds the battery:
forums.trossenrobotics.com/gallery/files/1/5/3/5/pb111440_thumb.jpg

As an option I also wanted to make a little head for some distance sensors. I’m using the new SES microbrackets for the head, and the head are mounted inside the body. The only modification needed to do is to remove one spacer in the front of the body. The head are mounted using a L-bracket that are screwed onto the hub that holds the right front leg. I drilled out some extra holes for holding the bracket, not really necessary but made it a bit stringer since the thicknes of the L-bracket and the body plate made the screw a bit short.

forums.trossenrobotics.com/gallery/files/1/5/3/5/pb111459_thumb.jpg

Here are some pictures with the head mounted:

A closeup picture of the head:

Thats it for now. I’ll post more info later.
Stay tuned… :wink:

2

Hi Zenta,

Happy to see you made your new project public! I’m always happy to see what you will come up with! But you already know that :wink:

I think you found a good way to solve the IK for the extra tars joint. I’m really excided to see how it will look doing body rotations. I think it will be pretty impressive!

I told you before that the Lipo attachment is simple but brilliant!

How did you made the body? It looks like a professional cut. Is it CNC made?

Thanks for sharing! I will follow your progress for sure! (as always)

Xan

3

Hi Xan!
Thanks for your comments!
You are correct about the body rotation, the extra joint sure helps. But there are some limits when it comes to large rotation values. The problem occour when the tars are getting passed the zero point/below the coxa joint. I guess you’ve seen it sometimes :wink: I’ve not solved the IK for a hexapod when the tars needs to go under the body. I did solve it for Felix though.

Believe it or not but the body are made using a very old (atleast 20 years) bosch jigsaw mounted upside down on a table. I got from my dad for a long time ago. If I get time I’ll post a picture of it. Oh I forgot to mention that I used 2mm sheet aluminium for the body. I guess 1,5 mm would also work (like LM used for Phoenix).

4

Hi Zenta,

That looks totally awsome :open_mouth: ! Great job. I could not imagine hand sawing those plates like that. I am prety much all thumbs and I am afraid I tried it, I would end up looking like another forum member :blush:

Hay Jim? Looks like another project for you! :astonished:

Let me know if/when you would like some XBEE code. I have taken a few detours and am trying to get a consistent set that works on Hex, Brat, Rover from DIY remote and PC…

Kurt

5

Yeah amazing! You really know how to work your tools. I’m pretty sure I can’t make such strait cuts by hand. You keep amazing me 8)

Xan

6

I used to make parts like that! I bought some adhesive backed paper. Designed and printed. Applied the paper to the material, then carefully cut out on the line. After cutting and drilling I would then scrape all the paper off the parts. 8)

Nice job on the chassis. I really like it. :smiley:

7

Hand cut panels? that’s insane! It looks like it was CNC water jet, at least that’s what I was thinking, but an old rusty jigsaw did the job? CRAZY!

Great looking design. Very cool. 8)

8

Very nice work indeed! You are to be commended, both for the design and the workmanship!

I must admit, I’m not seeing where the tars are remaining parallel to the ground. I did consider adding complaint (passive) feet to my “Little Cat” 'quad, but haven’t done so yet.

I did finally get LiPo batteries an a Turnigy 6A charger. Still figuring that one out. I did manage to get a charge into my 1600 mAH batteries. I’ve also bought a couple 3A BECs to go with the batteries. I’ll use two on my 'quad. Now All I have to do is get the new batteries and BECs (2 ea) into my 'quad design, and then swap out the '475 servos for '650 servos.

Again, congratulations on your new project!

Alan KM6VV

9

That thing is looking awesome. Your skills with a jigsaw are just ridiculous. :open_mouth: I can’t wait to see more of this project. :smiley:

Are you going to write Basic Atom code for the IK or use generated sequences?

Why do you have the spacers on the front coxa bracket?

I find it interesting that you have 3 consecutive DOF on the same plane. I don’t think I’ve seen a design like that used before. Very original.

10

Thanks for your comment Jim! Yeah, that method work very fine for me too. A bit hard for doing mass-production though… :wink:

11

Thanks for your comment Alan and Mike!

Thanks for your comment Kurt!
I’m getting very close for starting with the XBEE’s :wink: So if you have any updates in the near future I’m interested.
Thanks for sharing!

I’m using Xan’s v2.0 code for the BAP. And I’ve modified the IK part. So the code are made for the Basic Atom Pro.
I’m using the 1/4" spacer on the front coxa for increasing the distance to the body so that the front legs doesn’t get in conflict with the body. Actually **I wish there where a 1/2" spacer… hint hint Jim ** :wink: I tried the 1" spacer but that was just too much (didn’t look good either). But the 1/4" work fine though.

12

The pictures I posted yesterday was almost two weeks old… So I’ve been busy lately to make T-Hex fully assembled. As usual the wiring part took several evening/nights. I’m mounting the electronics in a sandwich in vertical position:

forums.trossenrobotics.com/gallery/files/1/5/3/5/t-hex_800600_20_thumb.jpg

This picture show the electronics mounted and I made a power switch panel ABS plastic. I’m thinking of mounting a XBee Explorer Regulated board beside the power switches. I’m using one switch for SSC32, BB2 and all servos.

This picture show the power wiring. Pay attention to that only one power source are used for both electronics and all 26 servos. The LiPo has a very high capacity and can deliver about 97 amps constant! and thats a factor for avoiding powerdrop when the all servos are under high load.

forums.trossenrobotics.com/gallery/files/1/5/3/5/pb151533_thumb.jpg

forums.trossenrobotics.com/gallery/files/1/5/3/5/pb151534_thumb.jpg

The new 5400 mAh Polyquest LiPo in place. This is a “relative” cheap LiPo compared to the one I use for A-Pod (5100 mAh). It can “only” deliver 18C instead of 25C, but as I mentioned above thats plenty of power for a robot.

forums.trossenrobotics.com/gallery/files/1/5/3/5/pb151535_thumb.jpg

Finaly, here are some samples of my fully assembled T-Hex:
(All my pictures can be seen at the TRC Gallery)

Two pictures that demonstrate full body rotation:
About 33 deg roll rotation:

And about 38 deg pitch rotation:

T-Hex and his little sister Phoenix:

T-Hex, A-Pod and Phoenix (my hexapod family) :wink: :

Single leg pose:

I’ll post a video when get time for it.

Thats all!

13

WOW! Awesome! T-Hex is huge!
I love it when it all comes together!

I like the shape of the massive legs besides the body. Did you use the single leg control (incl hold) to make the Single leg pose? :wink:

The electronics sandwich is a clever way to mount them. You don’t want to know how often I have to disassamble the BB2 to reach to the SSC for jumpersettings. I know, I know, I can also use a aditional switch for that. :stuck_out_tongue:

Just curious, Why are you using a separate power switch to switch the SSC logic?

Again, Great work!!

Xan

14

That looks awsome!

Nice tiny lipo :stuck_out_tongue: :laughing:

Kurt

15

Ha! Already on order. 8)

I will send you 6, one for each axis? Let me know…

16

So, uhh… do we have a Robot Designer of the Year award or something? :open_mouth:

No?

Well, here. You can have this then :bulb:. That’s all I’ve got. :stuck_out_tongue:

17

So the SES hole pattern at the top of the chassis. Is that where the Airsoft guns connect? :wink:

This is one of those defining moments in the history of the SES. Setting the bar a little higher for everyone else to strive for. Can’t wait to see the videos!

Again, nice work!

18

OMG! :open_mouth: The top of that thing is begging for a pan/tilt weapon of some kind! 8)

19

Whoa, Happy Birthday Zenta!!!

Here, have some delicious birthday cake! :open_mouth:

20

Hi Xan,

Yes I did use your single leg control and the hold function, it works great!

Easy access to both boards was one of the main reasons for the vertical sandwich. But it made the servo wiring a bit harder, had to extend the tibia and tars servo wires.

Good question. Since there was space for it, I just wanted to… LOL
I find it useful to only power the SSC when trying out sequences and calibration.

:laughing: Thanks James! To be honest I’ve already got alot of them… :wink:

Thanks Jim!
To be honest I didn’t have any special plans for the extra SES hole pattern, I just thought it would be useful for future ideas. It would sure be fun to try out some airsoft gun one day. Should be very possible to do.

You’ll all have to wait for the video. T-HEX walks very well, I’m only using the gait methods that uses the HalfLiftHeigth = 1, they work best for those large and heavy legs. I also want to work more on the code before posting a video. If I find time for it I’ll make a very short “teaser” video this or next weekend.

:laughing: Thanks! You are the first to notice (on the forum of course). So you changed the cake? :wink: Well, I love cake (also baking them). But this cake was just perfect for me (all of us I think) !

Thanks!