I am an electrical engineer and a professional model builder. I am looking into building a life-size working model of this robot.

The robot was donated from Jim Hensons Creature Shop in CA and being held at the California Science Center in LA, but two years ago, they could not give it away. I was told the main problems were, too big, too heavy, unsafe, and not working. After spending several months trying to find a new home for Big Blue, without any results, it was destroyed.

If I had worked on this robot and heard this news, it would make me sick. As an electrical engineer, I always keep my circuits schematics after I have completed a design, because you never know when you may need to use that same circuit again.

What I am looking for are: schematics, hi detail photos of the movie prop, plans for building the robot, CAD drawings, etc. I have tried for quite a while doing a Goggle Search and have not come up with much of anything.

Thank you very much,

JT

http://www.BuildingUltimateModels.com

Bots 4 Tots is a charity started this year to bring free robot building workshops to kids in the Chicago area, ages 10-15. Every penny of our donations go toward holding robotics workshops of no more than 5 children at a time, and our goal is to inspire some of these kids to pursue robotics beyond the workshops. Being that this is our first year, we're really looking to spread the word about our program, beyond Chicago and beyond the United States, to help bring this (our first) year's Summer and Winter workshops success so that we can continue year after year!

Check out our fabulous new website or this really cool article that BotJunkie just did on us if you want to know more. Email us at support@bots4tots.org. We'd really love to hear from you!

Thanks for your interest!

This forum will be much easier to use and will cover a much broader range of topics. From now on please post to the new forum :)

Many thanks, M@

I'm have just added a post on my new forum which gives some clues to the problem with servo linearity, you can read it here: http://www.hexapodrobot.com/forum/index.php

I will also be closing this forum soon, so please visit and join the new one. :)

Matt.

Thanks for the reply. It's good to know that it insn't impossible to use the SD200s, I know next to nothing about servos - Despite looking all over the place for info - so it is good to hear.

Yes i have got some sort of calibration scheme in place, actually i only just added it yesterday.

I tested each servo noting what pulse range they use for there full rotation (1900uS) and marked down what pulse length corrisponds to each 10 degree increment. Using that information i made something that'll send pulses a little more accurately. However i was a bit apprehensive before because the 10 degree increments seem to happen at random pulse lengths, and the centre position not where i would expect it to be - but was infact favouring one side of the pulse range.

It still can be upto a half degree off - as the pulse lengths that i marked down don't always hit the mark. But it is a lot better then before.

Thanks Again,

-Marty

Well I have used many different types of servo over teh years, and some of the cheaper servo's have very poor linearity, which makes doing anything usefull with them very difficult. Having said that, my first hexapod V1 was built with SD200 servo's, so you should be able to get something reasonable working.

I presume you have some kind of calibration scheme for each leg joint?

Matt.

I was wondering if anybody can help with a lil' problem i've been having. I decided to build a hex robot after seeing the fantastic Hexapods on this site, it's a wonderful hobby and up till now was going along pretty smoothly. However i'm having trouble controlling my servos correctly... My question is, Do different servos have varying degrees of accuracy, and does that mean that i'm going to have to buy some pretty expensive servos for this project? At the moment i'm using Ripmax SD200 servos but can't seem to find the right pulse width for the specific angles needed to move the legs properly.

I have developed and programmed a servo controller that lets me set PWM at 1uS intervals, meaning that i can set very specific pulse lengths. And i've worked out all the calculations for IK. So this is the last thing standing in they way before i can start experimenting with different gaits and all that fun stuff of a walking robot.

So are these Ripmax servos just not good enough for robotics like this, or am i missing something?

Thanks for reading.

-Marty

I'm currently finalizing the p.Brain-ds24 hardware design. I can add you to teh pBrain distribution list and send you a hardware datasheet if you require. Email me or use the contact form.

Cheers, M@

well, i ask that because i´m planning to construct an hexapod for my university and try to apply that control to many other things not only the hexapod, and i like to know if you have the controller for sale, or aprroximately for when you are going to have it for sale, or if you could help us to construct a similar p-brain..

thank you

I am just finalizing the hardware design for the p.Brain-ds24, if anybody would like to see the peliminary data sheet, please email me and I will send a copy. Feedback welcomed.

Matt.

I noticed on the 4b pictures (and others too I think) you directly connect the femur on both ends dirctly to the servo - I haven't gone through the whole forum but have you ever explained how you did it? I'm guessing you either made your own broach or found a supplier where you could just buy a broach and then you just broached the spline right into the aluminum? Comment?

--- hyperkinetic 

Hi, Yes I have been asked before, and it is somewhere here.. but not sure wher. The answer is that I was on a job where we could afford to have a bunch of spline punch tools made. While on the job I used the tool to spline the V4b kit, but have not used this technique on of my other hexapods.

Matt.

I noticed on the 4b pictures (and others too I think) you directly connect the femur on both ends dirctly to the servo - I haven't gone through the whole forum but have you ever explained how you did it? I'm guessing you either made your own broach or found a supplier where you could just buy a broach and then you just broached the spline right into the aluminum? Comment?

FIrstly, I don't use servopod, that would be a lynxmotion thang.. I think? I sue my own controller called the p.Brain.

The IK (Inverse Kinematics) is used to calculate the leg angles given a foot position in X,Y,Z, coordinates.

The cubic splines are used to move the foot from its current location A to a new location B via a point C between A and B. By using a spline to move the foot, a smooth trajectory is created with an ease in and out of the move.

Matt.

I have just about finished the router head, it took a bit of fiddling to get rid of the vibrations, but managed to find a small thrust bearing used on RC helicopters that did the trick. I have yet to mount it to the hexapod, plus there are a few tweaks to the spindle height, but it's more or less there.

You can see a picture of the router head in the development section.

The collet I have made currently takes 1/8" bits, but I need to finish the 3mm version as most of my small router bits are 3mm shank. The motor is a 14 turn 380 brush-less motor, which gives about 3600 rpm/volt, so plenty fast enough! It will run from 6 to 10 cells, I think I will add a heat sink, as it tends to get a little warm even under no load. The MUX servo moves the spindle 40/50mm up and down the guide rails to the work surface.

Looking forward to getting on the hexapod and running some g-code through it, but work is slowing me down.[/img]

1. Get one standard plastic wheel arm

2. Cut everything, only leave gear portation

3. make the plastic gear a little bigger than the hole on body

4. punch it with a hammar

--- Tony 

Hi,

Sorry, but i'm alittle lost..

What you're trying to say is:

1. u cut off the excess plastic portion for a servo horn. Leaving the gear portion

2. Drill a smaller hole in the leg

3. Press the pastic gear portion into the hole in the metal part?

** How does the plastic hold onto the metal part? Do you glue/ epoxy them? Since each part will be subjected to a large amount of force. They might just slip apart.

Cheers

Zack

In some respects I would hope not.. I don't mind people copying my designs, even offering advise and help as I have with Tony, but if people then give away / start selling the copy, I would find this somewhat irritating!

Matt. 

--- mdenton 

I would get pissed off as well. I'm just a student and it's a personal project. Just to let you know that I have absolutely NO intentions of mass production now or ever.

It's just that I don't wan to over/under design and end up wasting $$$ (18 servos is NOT cheap to a student). a HS645MG retails for about USD$62 in Singapore.. and 18 of these..

If there's any copyright issus, then it's ok. If not the dimensions, then perhaps the centre to centre length for each section (femur/ tibia) ?? At least then I can have a rough gauge of the max length I can stretch each part without having to worry of the motors not having enough torque

Cheers

Zack

M@

The picture will be removed soon, there will be no trouble so.

Zack

In some respects I would hope not.. I don't mind people copying my designs, even offering advise and help as I have with Tony, but if people then give away / start selling the copy, I would find this somewhat irritating!

Matt. 

1. Get one standard plastic wheel arm

2. Cut everything, only leave gear portation

3. make the plastic gear a little bigger than the hole on body

4. punch it with a hammar

I call it iDream. I'v finished body assembly today, below is the link to see the pictures. I did not put onboard PC today, too tired... I dare not test moves as it's droven by S3003, you are right, when S3003 lift the leg it always shake. is the same problem on 9206? I put a box under iDream to make it stand up, my poor servos can't keep hexapod stand when power off!! does iC can do this ? I am not sure if 3003 can do it with power on. But I am sure he can't if I put onboard PC on it's back.

Appreciate any suggestions, it's a clone of iC's body :-)

http://rainway.51.net/Picture/idream1.jpg

http://rainway.51.net/Picture/idream2.jpg

--- Tony 

Wow! It's almost a clone of the original that Matt built. Wonder if you'll release the cadd drawings/ dimensions?

Would like to ask how did you attach the tibia or femur to the servos? I don't see any servo horn or additional attachments.

Did you also made a servo punch to enable a direct coupling of a servo to the metal parts?

Cheers

Zack

So this is just a bit of fun, but has been in the back of my mind for some time now, and seeing one of Zenta's videos prompted me to give it a go.

Using B.F.Hexapod with an additional floating pen attachment, and a utility I wrote to convert DXF files into translation commands for my p.Brain controller, I have got the start of a walking CNC router! Why... I don't know.. it just seemed like a good idea at the time!

Watch the video here.

I still have to add a small routing head and test the cutting capabilities, but you get the idea. Resolution of the work area is pretty low, somewhere between 7 & 8 bit, with about 0.5mm repeatability, so it's never going to make precision parts! Also I have only implemented straight lines within the p.Brain, so the utility converts arcs and circles into segments.

One obvious improvement to resolution & repeatability is to have a work surface (in this case paper) that doesn't move around, in this video the only thing holding the paper in place.. is the hexapod's feet. In order to keep the router out of the way while walking, I plan on attaching a servo to the router head to swing the head from a horizontal to vertical position. Or maybe I will have come to my senses by then!! If I'm really mad I could try some 3D milling.. as technically the hexapod Is a 6 axis CNC machine.

Matt.

Well the foot contact area could be smaller, but this particular design using the gimbal and a passive adaptive foor works very well. It makes detecting ground contact easier, along with allowing the IK of the leg to work better. I designed a contact sensor that uses an FSR and a rubber ball at the tip of the foot, this arrangment works well if the foot is close to 90 degrees with the contact surface, but not at all at 45 degrees.

Matt.

Not too sure what you mean, but the foot is designed to adapt to different terrain shapes, and also stops the foot from sinking in to soft terrain. The sensor on this foot is a micro-switch attached to the gimbal. The foot base plate is mounted on a pivot fixed to the gimbal.

Matt.

--- mdenton 

Ops.. my bad.

What I meant was why are the contact area for each foot so big? If it is for distributing the robot's weight over a larger surface, then it is understandable. But is it really that much heavier?

What I foresee is the robot wouldn't be able to get a proper foot hold on gravels rocky surface..

Not too sure what you mean, but the foot is designed to adapt to different terrain shapes, and also stops the foot from sinking in to soft terrain. The sensor on this foot is a micro-switch attached to the gimbal. The foot base plate is mounted on a pivot fixed to the gimbal.

Matt.

Just a couple of questions:

The large surface (from the pics) seems flat and may not provide sufficient grip on the rough such as gravels. I am thinking like a micro-switch or FSR like lynxmotion would do the trick.

Is there any particular reason for using such a large foot area on each leg?

Cheers

Just added the developmet page back into the robotics section, with a design I have been playing with that uses 60mm size servos. Take a look here and let me know your thoughts.

M@

I call it iDream. I'v finished body assembly today, below is the link to see the pictures. I did not put onboard PC today, too tired... I dare not test moves as it's droven by S3003, you are right, when S3003 lift the leg it always shake. is the same problem on 9206? I put a box under iDream to make it stand up, my poor servos can't keep hexapod stand when power off!! does iC can do this ? I am not sure if 3003 can do it with power on. But I am sure he can't if I put onboard PC on it's back.

Appreciate any suggestions, it's a clone of iC's body :-)

http://rainway.51.net/Picture/idream1.jpg

http://rainway.51.net/Picture/idream2.jpg

There are several videos of various gait tests here:

Terrain Test 1

Terrain Test 2

Terrain Test 3

Terrain Test 4

Matt.

The majority of the weight is in the servos. For the V4b design, I use 12 x Futaba 9206 servos and 6 x Futaba 3002 servos. The 3002 servos are much smaller than the standard size servo, so I am probably saving weight with them. Also, the first two gears in a 9206 servo are plastic.

The battery I used would have been a Twicell NiMh 4.8V 2700mAh pack.

Matt.

For moving and lifting legs, the aluminium leg is very light so it's not problem, althrough the leg is longer than 12cm but 0.3KG is enough to lift a leg. But if it's for stand up, 3.5KG divided by 3 legs is 1.2KG, this would be a problem.

I think you have this the wrong way round. The Coxa joint needs the least power, and the Femur and Tibia joint need the most power. Don't forget that when you add a long arm to your servo, the torque is decreased.

Eg. if your servo has 5Kg/cm of torque, and your femur is 100mm long, the torque at the end of the femur will be 5/10 = 0.5Kg.

You do not need much power in the first servo (Coxa joint) unless you intend on going up steep hills or towing a trailer! ;) The 3003 should be fine for the Coxa, but I would use servo of at least 5Kg/cm for the femur and tibia joint.

Matt.

3 servos on each leg, one is coxa which link to body, another two for lifting calf and thigh. Lift calf and thigh is not a big job for iC as it won't real move the body, 3003 is working at 3.1KG torque so it's enough to do that job. the servo act as coxa takes the job to move the body, i guess you use 6x9206 as coxa :-) that's too expensive so i have 2 options:

1) Replace 6 coxa by 3010, it provide 6KG @ 6V and the price is about$35.

2) Use 3003 because when iC moves, it always has 3 legs lift up and another 3 move the body, so if my hexapod weight is around 5KG (with battery and mini-ITX), each coxa servo will take 1.65KG as average. it's just for horizonal move, not for lifting...

I did not test it yet as i am waiting for another 18 servos, god bless me. I need spend 1 week to assemble the hexapod as i need do some home work to make my plastic mounting bolt :), then the problem is how to programming and if my 4.8V, 2500mAH NiMH battery can or can not power 20 servos at same time.

iC uses 9206 and 3002 servos.

I'm not sure 3003 servos will give you enough torque!

Matt.

--- mdenton 

iC uses 9206 and 3002 servos.

I'm not sure 3003 servos will give you enough torque!

Matt.

http://www.tianyumodel.com/pic/yaokong/30031.jpg

I have no confidence with the hi-top servo, but i have to spend $200 to buy it :-) I can not go back so "good luck"

i did a test today to use a plastic gear cut from a rocker, then punch it to my aluminium arm, unfortunately it was too tight to made gear distortion. (use a hammer). the smaller gear can be easily installed but i am afraid it is not strong enough to move legs.

I need order 20 servos, and i have damaged my first servo yesterday. I don't know how to control pulse before read the instruction :-( i plan to damage one more to get lesson learned... Matt - pls. advise if simulative servo is good enough to move hexapod? I can't believe a digital one is triple than simulative.

How much did the parts cost in the end?

To avoid oxidisation, you can get the parts anodized.

Matt.

http://rainway.51.net/Picture/webupload1.jpg

Look forward to seeing them! :)