The Doll Forum

Hi all,

I'm interested in starting an open source animatronic humanoid robot project along the following lines:

Life sized
realistically articulating skeleton.
aiming ultimately for a maximum of powered joints, though to start with maybe just knees, hips, back, and some head movement
significant portions of the internal mechanics 3d printed, but not necessarily exclusively so where it doesn't make sense.
silicone skin cast with 3d printed moulds
appropriate control software, which at this point could be anything and isn't worth worrying much about until something exists to control
Not self powered. power would be delivered by an umbilical.

The emphasis to start with would be on the design files for the 3d printed parts, both internals and moulds.

The first problem i face in standardising on what software to use. For an open source project the really good stuff like solidworks or PTC creo are out due to cost. There is fusion360 which is popular in the 3dprinting world, but i'm suspicious of it's cloud model and i'm not sure it would integrate well with a github repo to allow collaboration alongside the other source files. Then there are things like Blender, loved in the 3d art community but not nearly so good for engineering CAD. The requirements here are slightly non-standard since we are dealing with a lot of organic shapes and trying to integrate that into more geometric geometries.

Source material must also be open source. Fortunately there is a great project, Makehuman http://www.makehumancommunity.org/conte ... loads.html which has us covered for the external models, and they are less buggy than models extracted form poser anyway.

Is there any interest in something like this?
Obviously this is a long term project, it might not happen at all or might just get part way, hopefully even just a collection of mould models would still be a useful resource for people from this community and many others.

Thanks

OSHR

This is a huge project, and Doll Sweet and Abyss have made a lot of progress on it. Of course they are all costing a premium, but those robot dolls are coming this year. It will take a lot of time and effort to catch up with those guys.

I think you have to start with the type of 3D printer required, how the doll will be designed, how will files be converted, etc., etc., etc., ...

This is almost like a Project Apollo.

Personally, I strongly feel that dolls should be as completely pneumatic and soft robotics as possible.

I've done some preliminary research, and it's (soft robotics/pneumatics) a better bet than trying to make a traditional robot (metal and servos).

Quite frankly you would have to develop most of the required designs, tooling, controls, from scratch which is THE challenge.

However, it would be worth it.

Also, don't forget the 'pipeline'. Going from avatar to VR to doll to android is the way to do it.

I agree with LDF that soft robotics is a better solution. It was already years ago that I watched a demo of artificial muscles contracting from electric input. I think the first mistake is trying to use motors to move joints. The sound is always bad and it's very hard to have smooth movement. Everyone is so jerky right now.

Sure, many of us are interested in robotic dolls.
Trying to do something so complex on a 1st run is another thing though.

Better figure out how to make a top quality doll first. Abyss has been working on it for.... 10+ yrs?
And is just now getting into robotic heads.
Now if you don't have a 9-5 to go to and have a spare $50K+ to get started, great.

Forget 3D printed molds. Professional molds are clam shell design, tool coated fiberglass.

And anything so complex will probably need a removable mesh reinforced silicone skin. The skin will probably still need to be molded somehow.
Silicone dolls have the skeleton and foam core suspended inside the mold and silicone is poured in through the neck hole. Any internal problems with even these simple dolls require cutting the silicone open to make repairs then having to glue them back together. This is a no-go for a robot where you will have 100's of things that may need serviced.

3D printed dolls/androids wouldn't require a mold although 3D printing is still rather slow. Sculpting, mold making, and casting dolls is an art/science.

Could you 3D print the inner android and then cast the outer androids 'skin'? Sure.

But, you can 3D print an entire doll/android w/ a multi-extruder head.

You would, however, need to literally design/make your own 3D modeling app, design/fabricate your own 3D printing platform, etc. .

For example, if it's an open sourced doll, just how open is that? Bring your own controller boards?

So, I'd love to see a 3D printer and design app that can do the job.

However, we're starting at ground zero. I haven't sen a fully integrated 3D printed pneumatic muscle, for example.

Thanks for the replies.

So some further points about where i see this going:

I too had at first thought soft robotics was the answer but unfortunately it's just not. Pneumatics is out of the question since air compressors are bulky and extremely noisy. It is also very hard to servo control pneumatic air-muscels. not to say it can't be done, but it's really hard. I had initially thought some sort of hydraulic equivalent to the air-muscel might be the way to go, but the difficulties in fabricating a pump that can deal with a water-based working fluid along with all the valves are just too much. trying to pursue the soft-robotc route turns the project into one of inventing the necessary components, rather than using off the shelf parts. It's just not practical unfortunately.

That brings us on to motors. The noise associated with motors is mostly the gearbox and gearing is not really suited to this project where limbs need to be able to be 'driven' manually when the robot is powered off, and take huge forces when it is operating. some compliance is needed. I propose to use high powered stepper motors, driven by DSP controllers to keep the noise down. I have some ideas for getting some mechanical advantage without an actual grarbox where it is absolutely necessary


The 3dprinting of the mould, i agree is sub-optimal but it allows people to contribute to the project and pick the bits they are interested in. with a requirement on a full-body profesional fiberglass mould . that wouldn't really be possible.

The hardest part of this thing is the skin. Yes it would need some reinforcement, i was wondering about the possibility of adding short kevlar fibers to the silicone rather than having to actually cast a fabric mesh into it. making the skin fit around the internals without leaving unnatural bulges or holes where some padding isn't quite right is a significant challenge. I understand they cast the silicone around the frame and foam filling for the pro dolls and have always wondered how they keep the frame/filling located in the center of the mould. This would be a harder problem where there are moving parts that need to not get gummed up. The whole mechanical internals would basically have to be in one big bag to keep the silicone out.

LDF wrote: Personally, I strongly feel that dolls should be as completely pneumatic and soft robotics as possible.

I've done some preliminary research, and it's (soft robotics/pneumatics) a better bet than trying to make a traditional robot (metal and servos).

LDF: I've done some pneumatic animatronics installations but have not heard of soft robotics. Do you have a link or two that would be a good start for me to see what's going on with that as it might apply to sex dolls?

All: Due to the complexity and cost curve, I would be more confident getting involved in a smaller project. For example: Vag inserts that offer a higher function of stimulation than just shape and heat. I don't know-- rotating tip? Now we're talking about something beyond what a human sexual partner could perform.

In the early days of synthesizer keyboards, there was a push to have them make them emulate live orchestral instruments. After it crossed they crossed that uncanny valley, there finally came a return to embrace of their synthetic nature and that's when their unique possibilities really opened up for EDM and other kinds of experimental electronic music. I hope to see the same with dolls. What will they look like in 35 years? Human... or better?

3D printing can certainly be used to make models or molds for a lot of the doll.
I'm thinking a robotic doll will be built completely different than a common silicone sex doll.
As mentioned above, a silicone sex doll is basically skeleton & foam inside a poured silicone body.

A robotic doll will probably require skeleton then robotics attached. Individual silicone muscle and fat packs will have to be molded and fit in and around the skeleton and robotics then attached somehow. Then a tight fitted silicone skin pulled up over all that. Back zip? Glued around the waist line?

Now maybe if you have a robotic doll w/ an internal structure and robotic system that is 99% unbreakable, you could do a liquid pour around that for a permanent skin. But if the robot suffers many failures, people are not going to want to keep cutting open and gluing shut their doll.

SuperNothing wrote:

LDF wrote: Personally, I strongly feel that dolls should be as completely pneumatic and soft robotics as possible.

I've done some preliminary research, and it's (soft robotics/pneumatics) a better bet than trying to make a traditional robot (metal and servos).

LDF: I've done some pneumatic animatronics installations but have not heard of soft robotics. Do you have a link or two that would be a good start for me to see what's going on with that as it might apply to sex dolls?
All: Due to the complexity and cost curve, I would be more confident getting involved in a smaller project. For example: Vag inserts that offer a higher function of stimulation than just shape and heat. I don't know-- rotating tip? Now we're talking about something beyond what a human sexual partner could perform.

https://biodesign.seas.harvard.edu/soft-robotics

I am actually seeing the possibility of building a 3D printed dual-axis pneumatic servo (shoulder/hip joint, etc.).

Just for the record you could always 3D print and then assemble components into a skeleton with 3D printed dual axis pneumatic joints for example.

IMHO, you can already 3D print silicone and TPE, so the rest is doable.

I could go on and on, but we're still at the spit balling stage.

Make the robot doll hard plastic and then make the sex holes such as anus, vagina, mouth out of silicone.

LDF wrote:
https://biodesign.seas.harvard.edu/soft-robotics

That is inspiring! Imagine that kind of technology wrapped around you similar to a woman adept at her kegel exercises!

Tybalt wrote:Make the robot doll hard plastic and then make the sex holes such as anus, vagina, mouth out of silicone.

Could be hard plastic shell with plastic articulated limbs having a layer of firm foam and silicone skin.

Speaking of spitballing. Very few know that the 3D printer was invented by a guy in Greenville South Carolina. He came up with the idea from spitting off of the top of a building. He imagined that if he could do that enough he could eventually construct a 3D model by layering the spitballs one on top of each other in a specific pattern. So as you see spitballing can actually turn into something extremely cool.