Author Topic: Our hexapod Spider P.I.G (Read 9173 times)

Fredrik Andersson · « **on:** January 05, 2008, 04:55:15 PM »

Blah! I'm copying the my whole posts from the Lynxmotion community:

I'm working on a hexapod as a school project. We call it Spider P.I.G! We've just started constructing it (We spent this whole semester on planning, funding and ordering parts). However, before We get any further, I thought it would be great to hear some of Your suggestions if there is something we planned that really won't make it.

Here's a couple of CAD images (sorry it's so sketchy, I'm no good with CAD

):i

We will use S03T GWS servos though i ordered 6 of each of the different speeds. Thought it was a great idea when ordering, especially for the innermost joint where We'll put the fastest servos. I'm not sure anymore though.

We already made the torso (may i call it that on a hexapod?) and here is the result (I'm pretty happy of the result considering the tools we were restricted to):

So if something is in need of change it can't be the torso, cause We're not remaking it!

It took a whole lot of time.

[edit]The torso may look thin and fragile, but theres more to it. There will be a body of several layers of styrofoam on top of this piece (and there will be layers of plastic in between the styrofoam so that we can screw everything together).

Also, I plan on making some kind of "spider silk" for the robot. By that i mean tape ^^ Now wouldn't that be cool? It will have to roll out a strip of tape before it arrives to an edge. Then it can easily hurl over the edge and safely roll down through the tape to the floor. Not sure how it will be done exactly, but I have lot's of partly ideas for how it will be done.

The latest post:

I have made lots of progress lately. Here's some pictures (I hate my camera

):

This was before i turned the coxa upside down. I had to do that because there will be another layer to stabilize it:

This is a blurry image of the coxa when it's turned upside down:

I added some improvised hinges for the layer that will stabilize the coxa (the white stuff):

The tibias added. If anyone need to make some fast and cheap tibias, just use a wooden stick, flatten one side 3 cm in on the stick to make it fit more to the servos flat surface and hot glue it onto the servo. Results in very much rigidity. They are somewhat temporary though. Only to complete the torso left now:

She is pretty huge. From one toe tip to the opposite one, it is 74 cm (29.5 in):

Next, I'll continue on the torso. But I'll probably wait with that until i can work with my teammates.

Rebelgium · « **Reply #1 on:** January 05, 2008, 05:41:18 PM »

Looking good!

Tnx for sharing and keep it up.

airman00 · « **Reply #2 on:** January 05, 2008, 07:09:19 PM »

One word: WOW!

ddemarco5 · « **Reply #3 on:** January 05, 2008, 07:43:55 PM »

Looks great, can't wait to see finished product.

Ro-Bot-X · « **Reply #4 on:** January 06, 2008, 02:06:33 AM »

Nice job! I am almost sorry that I didn't build one too!

I have a few questions about the electronics.
But let me tell you about what I want to build and see if it is remotely similar. I want to build a biped (that's the only thing different I guess) and use a Lynxmotion SSC-32 (actually custom made SSC-24) to drive the servos. And I want to use the AVRcam to find objects. The main MCU (not sure yet which will that be, thinking of atmega 32 - since I have one, or atmega128) will have to talk on the serial UART bus at 115200baud to both AVRcam and SSC-32. The main MCU will have to do the mapping and A.I. decisions.
Now the questions: How are you controlling the servos? Will you have a different MCU for logic decisions? Will you include the AVRcam in this project?

I wish I could start on my project soon, but it looks like I'll have to wait untill I'll be in Canada. Mean while I am working on a wheeled platform to be able to work on the code.

Crooksie · « **Reply #5 on:** January 06, 2008, 07:06:10 AM »

That's pretty amazing.

Please post a video when you get it moving, I'm intrested to see how it will move.

Fredrik Andersson · « **Reply #6 on:** January 06, 2008, 02:34:48 PM »

Thanks guys, i always feel so empowered and motivated hearing positive feedback on stuff

Ro-Bot-X:
I haven't come far enough that i can experiment with that. Haven't been thinking about it yet either. I'll deal with that when the body is done. I will use a Arduino for controlling it, but first i will experiment with gaits and such from a computer.

I will not use a AVRcam though. The sensors it will have is 6 IR rangefinders on top of it, slightly angled downward so it can see edges like stairs and such.

HDL_CinC_Dragon · « **Reply #7 on:** January 06, 2008, 02:37:53 PM »

Hexapods have to be hard as hell to get the gaits right. More power to ya for going for it

Any idea of how durable it will be when its done? Basically how resistant is it to physical body or circuit damage? perhaps an inch or two thick diamond plating all around would help! But then of course you would need like 100kg/in torque servos lol

Fredrik Andersson · « **Reply #8 on:** January 09, 2008, 01:01:57 PM »

I just wrapped together a beta of the first utility for Spider P.I.G. We will for example create an open source sequencer for servo controllers (yes,it will support custom controllers) in java.

It's calculating the angles of the joints to reach a certain position (This is what's called reverse kinematics, right? No... wait a minute, it's forward kinematics, isn't it?

).

If you are curious about how it's calculated, the source code is distributed and you can find the calculations in the paint() method. Note: This calculation only works if the joints are the same length, which is the case on Spider PIG.

paulstreats · « **Reply #9 on:** January 10, 2008, 07:23:09 AM »

Nice applet.

How many steps of motion are you planning on using for each complete leg motion?

Are you planning on porting the code directly over to microcontrollers for on board control?

Fredrik Andersson · « **Reply #10 on:** January 11, 2008, 01:05:59 PM »

I need to say that we tested the fully assembled body today and experienced major failures of the coxa (the innermost joint of the leg). The material is too thin and will bend, thus making the legs completely crooked. The other parts hold really well though. We will have to beef up the construction. Just so you know, if you intend to use our design.

HDL_CinC_Dragon · « **Reply #11 on:** January 12, 2008, 11:52:35 AM »

dont suppose using some small Aluminum square stock could be an option?

Admin · « **Reply #12 on:** January 13, 2008, 01:45:45 PM »

Quote

It's calculating the angles of the joints to reach a certain position (This is what's called reverse kinematics, right? No... wait a minute, it's forward kinematics, isn't it?

you mean inverse kinematics?

I'm a bit worried about torque problems . . . note that you want to keep all the weight as close to the legs as possible and far away from the robot center. This reduces required torque, because the length is reduced from the actuator. Torque is force * length. Consider this for battery placement . . .

Quote

dont suppose using some small Aluminum square stock could be an option?

good idea. bend the aluminum into two 90 degree angles, something like a C shape.

hazzer123 · « **Reply #13 on:** January 13, 2008, 02:04:40 PM »

Hmmm if the batteries are placed nearer to a leg of a robot, then it will be further away from a leg at the other end of the robot and therefore more torque is needed.
From this it seems that the centre of the robot would be the best place to put the battery since it stresses all legs equally, spreading the work out.

gamefreak · « **Reply #14 on:** January 13, 2008, 02:12:53 PM »

you could create a custom battery pack, with a battery(or two) at/on each leg

hazzer123 · « **Reply #15 on:** January 13, 2008, 04:30:09 PM »

Which would effectively be equal to all of them being in the centre, wouldn't it? The centre of gravity is all that matters, at least that is what I'm thinking.

Admin · « **Reply #16 on:** January 13, 2008, 04:44:30 PM »

Quote

Hmmm if the batteries are placed nearer to a leg of a robot, then it will be further away from a leg at the other end of the robot and therefore more torque is needed.

I was envisioning the weight to be balanced with other components.

Or you can always take that 2000mAh battery and instead use two 1000mAh batteries in parallel to balance out that weight.

Many options . . .

SmAsH · « **Reply #17 on:** January 13, 2008, 08:08:08 PM »

hmmm yes i think you should find a battery pack that would fit on the leg and have one on each...anyways just an idea

Rebelgium · « **Reply #18 on:** January 14, 2008, 07:43:12 AM »

Quote from: SmAsH on January 13, 2008, 08:08:08 PM

hmmm yes i think you should find a battery pack that would fit on the leg and have one on each...anyways just an idea

I disagree, then you'd have to buy more powerfull servo's to carry the extra weight...
Just place the battery pack in the center, and as low as possible.

SmAsH · « **Reply #19 on:** January 15, 2008, 05:25:48 AM »

hmmm....yes but youll need a very powerful battery and it wont run for very long...

Fredrik Andersson · « **Reply #20 on:** January 26, 2008, 11:40:01 AM »

Quote from: SmAsH on January 15, 2008, 05:25:48 AM

hmmm....yes but youll need a very powerful battery and it wont run for very long...

It ran nicely on a 6v 1250mAh battery (but don't get too exited, we haven't managed to get it walk yet, just tested the stability in one pose). Don't know for how long it will do so though. I have another one that can be connected in parallel anyway, so that shouldn't be a problem.

So, we have started fixing the failures at the construction. The fixes is made out of wood, which may seem a little ghetto. But they do their job really well. The only problem with it being constructed out of wood is that it wear out much more easily, but as we don't have a need of attaching and detaching them anytime there won't be any wear on it.

Ugh, hating my camera. Anyway, the L-shaped brown structure is the fixing part.

hazzer123 · « **Reply #21 on:** January 26, 2008, 11:57:19 AM »

60V! I assume you mean 6!?!

Nice work though man.

Fredrik Andersson · « **Reply #22 on:** January 26, 2008, 12:26:46 PM »

Quote from: hazzer123 on January 26, 2008, 11:57:19 AM

60V! I assume you mean 6!?!

Nice work though man.

Lol, yeah, of course i meant 6v

Thank you!

[edit] I continued to work on the robot this evening and it was fun!

While the whole robot is stored at school I snatched of a leg to work on the stabilizing piece over the weekend. But i finished it so fast and got bored after that, so i realized I still got my arduino at home. I thought i would experiment with that one leg.

I programmed a code that takes x and y positions from a serial terminal. It calculates them into angles instead, just like the java applet i made (It was based on that code actually). Also, i added a function that made a circular movement of the leg tip with a changeable radius.

It ran really smoothly. However, i thought the movement looked a little oval, so i decided to look up if it really was so. I added a pen at the tip of the leg so it would draw the route. I was right! The movement was a little oval. Please note that the servos weren't really calibrated.

Next up, I'll make the movements 3D capable. I haven't touched the servo rotating horizontally yet.

Fredrik Andersson · « **Reply #23 on:** February 08, 2008, 04:34:02 PM »

Got a chance to test and see if my EEE PC would fit on out robot. This is kind of how it would look if i attached it to the robot:

Rebelgium · « **Reply #24 on:** February 08, 2008, 05:50:08 PM »

Looks so cool!
Can your servo's handle the weight?

Half Shell · « **Reply #25 on:** February 08, 2008, 07:16:25 PM »

The EEE pc is just 2 pounds. Only slightly heavier than what normal electronics for this would cost.

Glad to see someone beat me to the punch of making an EEE compatible robot :-D.Are you using the EEE to control it? What OS on the EEE?

Your fellow EEE owner,

Half Shell

Rebelgium · « **Reply #26 on:** February 08, 2008, 07:37:41 PM »

Quote from: Half Shell on February 08, 2008, 07:16:25 PM

The EEE pc is just 2 pounds. Only slightly heavier than what normal electronics for this would cost.

Glad to see someone beat me to the punch of making an EEE compatible robot :-D.Are you using the EEE to control it? What OS on the EEE?

Your fellow EEE owner,

Half Shell

He only placed the EEE on top of his hexapod, it isn't attached nor connected.

Fredrik Andersson · « **Reply #27 on:** February 12, 2008, 12:24:16 PM »

Half Shell: Yes, like Rebelgium said, i only placed the EEE PC on top of my robot to see what it would look like and if it would fit well. But it should be fairly easy to do it i guess. I just have to hope the servos ar strong enough, our hexapod is already pretty large compared to most of what I've seen (though it isn't very heavy compared to the size). It will take a while before i can integrate it with the EEE PC so i think you will be first to it anyway

BUT, i will be the first one to make a hexapod with it! (you didn't intend to build a hexapod did you?)

Anyway, i run eeeXubuntu on the EEE PC. I like it very much.

Also, i've continued on my IK utilities for the robot. Here's a couple pics of it:

Right now you can only control each leg manually, or moving the body compared to the legs manually. But it's work in progress and I'm counting on having it done by the end of this week.

The IK calculations are all right (but one thing i haven't taken account of yet is the length of the coxa. I figured that could bet the reason the circles drawn by the leg was a little misshaped).

Trumpkin · « **Reply #28 on:** February 12, 2008, 01:56:59 PM »

coolness. I'm planning on making a humanoid robot for my next project.

Fredrik Andersson · « **Reply #29 on:** February 18, 2008, 04:00:44 PM »

Finally i have a video to show. It's not much though. I hope I'll be able to throw together a better one by tomorrow with a better camera. Until then, enjoy this:
[youtube]http://www.youtube.com/watch?v=ScpKr6P5DMI[/youtube]

This is quite a large step for me even though the robot can't walk yet. I'm not used to Java, so I had a really hard time getting the serial communication to the servo controller to work. Not that i have this fixed, it will be easy to focus on creating a walking gait.

News:

Author Topic: Our hexapod Spider P.I.G (Read 9173 times)

Related Topics