One more robotic 3D-printer/assembler

[Dan_2013]

Hello!


I want to make 3D-printer/mechanic assembler, which will be able to print parts from plastic and to process them automatically (to separate printed part from platform, then placing it in a tray and preparing printing platform for printing next part) and to assembly from them more complicated parts or finished items (for example small robots, or 3D-printer details). Check: Publix ad this week and Kroger ad this week on Weekly Ads.

At first I simply wanted to attach to the printer few robotic arms, but recently I saw this video -



As I can see the robotic arm can serve as a 3D-printer itself.

Besides, if to add machine vision to such robot, which will allow to him to determine accurately 3D position of objects in operational and neighbour area, and to add a good callback from robotic arm sensors, then it will be possible to create easy-to-assemble self-calibrating printer. (The maschine vision applications now allow to produce 3D shape of surrounding as a points cloud, or as ready 3D model, but this applications need binocular camera for determining Z-coord, for example - https://www.leapmotion.com/. This device is pretty accurate. Actually this is simple analog of industrial optic 3D scanners, which are used in quality control solutions - http://www.aicon3d.com/en/produkte/b...f-einen-blick/ ). So, the easy assembling and self-calibrating 3D-printer (by mean of software, 3D-optical and callback sensors) will be very convinient thing as well.

Well, I hope that your robotic arms may serve as robotic 3D-printer, but the question is - will it be possible to obtain enougth accuracy for this purpouse with current models of robotic arms?

On yours videos it is visible that manipulators under loading is noticeable displaced because of their not very big size, but it would be possible to solve this problem, having used for holding of a printing head not one manipulator, but three or four, as in delta printers:


http://www.flickr.com/photos/5808532...n/photostream/



The question is will robotic arms able to provide accuracy for printing with resolution 300-500 microns? This resolution will be enougth for a wide range of constructing tasks.

Actually thats not a problem if needed accuracy can not be reached, because in this case it will be enought just to combine robotic arms with current schemas of 3D-printers. But if this accuracy is available it will be possible to use this robotic printer also as a small CNC maschine tool with different heads, which can serve itself in transitions between different operations (changing heads, bits and so on). For example it will be possible to use Dremel-like head and a wide variety of it's bits to drill, cut, polishing (and so on) parts from plastic and other materials (like here - http://fablabs.ru/wiki/index.php/Кулибин_1212)
Or use heating tool, or painting (small aerograph for example). And in addition to this the robotic printer will have more DOF than current 3Dprinter heads. This will be usefull even for printing head, not for painting and drilling only. For example in this video -



printer works almost fine in any positions (extruded filament came unstuck a little because of shaking, not because of rotation). So i think it may be possible to make additional prints on existing parts under different angles. This may be usefull option.

And, as final solution, it is possible to change constructon of robotic arm, like in hydravlic or pneumatic manipulators,



and to connect manipulator parts not directly to stepping motor axis, but through the some gears and rods to increase their accuracy and strength with remaining their not very big size (like connections between printing head and stepping motors in current 3D-printers). But first it would be desirable to learn what accuracy can provide existing models of robotic arms like PhantomX Pincher for example.

The super-goal of this project's sketch may look like this - to make robotic 3D-printer which will be able to assembly it's own copy from existing universal modules (like in Lego MindStorm) which will include controllers, sensors, motors and connectors, and from set of parts printed by themself. But even in not complete implementation this will be very usefull tool for hobby constructors, technologists and lazy coders like me

Thanks.


P.S. I apologise for not very clear explanation, english is not my native language and I've composed this message with help of electronic translator.

[CasperH]

Awesome! Sounds like an interesting project to follow. Do you have any designs ready (3D or notebook sketches)?

When you said print and part manipulation, I initially thought of the thing-o-matic, have you checked that out? It ejects the printed part after the print to continue on the next one. http://en.wikipedia.org/wiki/MakerBo...#Thing-O-Matic

Something from a while ago for me, at school we had a project with 3 industrial robots in a sort of assembly line. Basically we had a simple fill/storage mechanism and the robots built the train, loaded it and controlled the movements. The main reason I bring this up is that the parts that were assembled are Lego, which is from ABS, which is something you can print. I have pictures/video if you are interested.

Are you considering this as a hobby project or something for school/work?

[Dan_2013]

Do you have any designs ready (3D or notebook sketches)?

Just painted. Something like this. It may have 4 more manipulators/or wheels on lower level for moving/manipulating with support for parts.

When you said print and part manipulation, I initially thought of the thing-o-matic, have you checked that out? It ejects the printed part after the print to continue on the next one. http://en.wikipedia.org/wiki/MakerBo...#Thing-O-Matic

No, didn't heard about this. Thanks a lot for link. Will look at this printer. I'm using 3dUP printer and it's always a headache with it's supporting layers and removing and cleaning perforated platfom. And even with all these stuff large details from ABS printing with deformations.

Something from a while ago for me, at school we had a project with 3 industrial robots in a sort of assembly line. Basically we had a simple fill/storage mechanism and the robots built the train, loaded it and controlled the movements. The main reason I bring this up is that the parts that were assembled are Lego, which is from ABS, which is something you can print. I have pictures/video if you are interested.

Are you considering this as a hobby project or something for school/work?

Yes, of course its intresting. All usefull videos and links are welcome. I think it's very simple idea and it's strange that it is not implemented yet as personal device. If you like this idea too and can help with it's implementation - it's really great. Or may be you know better variants. The goal is just make a presonal assembler for mechanical parts or creating circuit plates and so on. Unfortunately I'm lamer in technical things. For me all this robotic stuff is "secondary" hobby. "Primary" was RC planes and 3D printer helps me build different things for this. And after this I'm starting to learn 3D-printing and robots.

Few time ago the robots was not very functional (they was usefull in strctly determined and specialized operations and applications) because they was almost blind, deaf and so on. But now with speed controllers (in comparison with older ones), with technical vision and with all other sensors, they should became something like their images in fantastic books. It's very intresting

[CasperH]

Just painted. Something like this. It may have 4 more manipulators/or wheels on lower level for moving/manipulating with support for parts.

Wow, that looks like an operating table from a sci fi movie. Do you have any specific type of product or object in mind what you would like to fabricate? Why have that many heads? I can imagine you can get a long way with just 2 (milling and printing) or expand it a bit with a modular head setup.

Yes, of course its interesting. All useful videos and links are welcome. I think it's very simple idea and it's strange that it is not implemented yet as personal device. If you like this idea too and can help with it's implementation - it's really great. Or may be you know better variants. The goal is just make a personal assembler for mechanical parts or creating circuit plates and so on. Unfortunately I'm lamer in technical things. For me all this robotic stuff is "secondary" hobby. "Primary" was RC planes and 3D printer helps me build different things for this. And after this I'm starting to learn 3D-printing and robots.

Ok here goes, I did two projects at school that you may like. Unfortunately my blog is in Dutch so Ill translate the assignment for you so you can get the idea.

The first assignment was to use three industrial robots to assemble something. We had three groups, 2 guys per robot, in the end we combined all the three robots in to one assembly line. As I mentioned we used Lego, a train was assembled, loaded and a third robot would put the train back on the track back and also controlled the velocity. See video:



The second assignment, which is probably even more up your ally. The assignment was to come up with some product that had moving parts which had to be made through CNC fabricating on a lathe and mill and had to be assembled by an industrial robot. We came up with an adhesive tape holder, that when turned, would drive a v-shape motor with pistons that are clearly visible through the transparent material. The robot was an industrial PUMA. Here is a video:



As for helping out, I currently have my own robot project and recently got myself a delta robot 3D printer that are taking up all my time. If you have questions here on the forum, I am happy to answer if I can. If you are new to robotics, in my humble opinion you have chosen a very ambitious way to get started. Are you planning on doing everything from scratch or are you more looking at combining existing parts, kits and things?

Few time ago the robots was not very functional (they was useful in strictly determined and specialized operations and applications) because they was almost blind, deaf and so on. But now with speed controllers (in comparison with older ones), with technical vision and with all other sensors, they should became something like their images in fantastic books. It's very interesting

Yes, its a great time in history to be a robot hobbyist. Lots of available things, resources and great communities with like minded people (like the Trossen forum ).

[jwatte]

One problem with arms is that the backlash of the mechanism multiplies out the entire length of the arm, making for relatively poor precision. Gantry-style printers have an absolute amount of backlash, as there's no mulitplication by distance involved.

There *are* high-quality arm-based CNC mills, doing 5DOF milling by moving around objects. They are very expensive, though, because of the engineering needed to overcome the above, while carrying a heavy spindle.
For an example, check this out: http://www.youtube.com/watch?v=RnIvhlKT7SY

[Dan_2013]

Wow, that looks like an operating table from a sci fi movie.

Yes, exactly This is very common and simple idea, especially in sci fi movies. But I see it as overturned quadropod or hexapod

Do you have any specific type of product or object in mind what you would like to fabricate?

Robots and it's parts. Different robots - flying, swiming, driving, stepping. Printing, milling, soldering and than assembling from produced and preproduced parts in automatic mode. It will be very intresting task itself - to create not only blueprints for parts and STL models, but all technological process for producing these things. And also for using it as micromanipulator for electronics in automatic mode or in handly mode with joystik, zoomed video on monitor and different type of manipulators; two hands often not enougth when soldering something And for assembling small details and different small stuff it also will be usefull. It is possible to print different grasps for special tasks even for one detail or operation and than recycle them. And I think these are a lot of applications which I've not enumerated.

Why have that many heads? I can imagine you can get a long way with just 2 (milling and printing) or expand it a bit with a modular head setup.

Well, the schema above mean that robotic hands will be used as delta-manipulator all together to decrease backlash as much as possible. So the heads will be movable part which will be captured by manipulators by their corners, and after using placed back to holder. Printing head and filament moving mechanism is not very difficult part, the milling head will be also not very difficult, it will work with soft materials. I think may be usefull hot air gun head and painting head. Also it may be used laser cut head, but under protective screen. For 3 DOF table all this heads mounting very easy. For robotic assembler it will be more difficult, but all these tools will not work with hard materials, so they will not very heavy.

The first assignment was to use three industrial robots to assemble something.

Very nice.

The second assignment, which is probably even more up your ally.

Yes, exactly. Wow, thats realy cool This is almost what I mean, but with two additions: 1) specialised graps, printed for developed assembling operations, and 2) callback with 3D realtime optical scanning of operational area. And all operations runs in a much smaller space. Did you use optical 3D scanners in this solution?

As for helping out, I currently have my own robot project and recently got myself a delta robot 3D printer that are taking up all my time. If you have questions here on the forum, I am happy to answer if I can.

Yes, I'm planning to make robotic arms to act as delta printer branches, so I'll be thankfull for help, if I'll face serious difficulties.

If you are new to robotics, in my humble opinion you have chosen a very ambitious way to get started.

That's Ok, It's not a first time

Are you planning on doing everything from scratch or are you more looking at combining existing parts, kits and things?

I'm nub in mechanics and electronics, but I'm not the worst programmer. I will use existing parts and try to make them work as I need. Also I'll use printed mechanic parts and may be purchased mechanic parts. First I'll start to experiment with robotic arm, if there are no problems with delivery. First task - to check current available accuracy and find solution to get needed level of accuracy (for example just use instead of arms rails like in Bukito 3D printer with rotation in the base of vertical rail). I hope that key solution in this task will be realtime optical 3D scanners - for run-time correction of manipulator movement. May be I'm wrong. This thing need to be checked second.

Yes, its a great time in history to be a robot hobbyist. Lots of available things, resources and great communities with like minded people (like the Trossen forum ).

Yes, as I see this place is free from trolles and casual people. Very good. I'm glad that I'm here

[Dan_2013]

One problem with arms is that the backlash of the mechanism multiplies out the entire length of the arm, making for relatively poor precision. Gantry-style printers have an absolute amount of backlash, as there's no mulitplication by distance involved.

Yes, this is the main problem for now. It will be pretty cool if it will be possible to use robotic arms both as printer and manipulator. I think it is possible to minimize the backlash in arms by blocking most part of movable mechanical connections (by just stopping the motors, or with help of some kind of mechanic locks) and using arms as delta-printer branches in printing and precision mode. May be this will be enougth, in other case I'll try to find another solution. My goal - on-table personal solution for hobby purpouses and may be for some small home tasks.

There *are* high-quality arm-based CNC mills, doing 5DOF milling by moving around objects. They are very expensive, though, because of the engineering needed to overcome the above, while carrying a heavy spindle.
For an example, check this out: http://www.youtube.com/watch?v=RnIvhlKT7SY

Yes, this is really state of the art technologies. The mechanical problems with precision positioning under heavy load was successfully solved by engeneers a lot time ago. Now we have one more key element - 3D machine vision. In hardware part it's simple two cameras and projector with determined light pattern. Cameras analysing pattern deformation on the picture and builds 3D points cloud with high resolution and very good accuracy. For example for end-user applications developer now using only two cameras, even without pattern projector (https://www.leapmotion.com/ I've asked project team on forum is it possible to use their controller as 3D optical scanner, but they are not answered, looks like they just tired to answer this question for now option which allow to get points cloud form leap controller not enabled, available only 3D coords for thingertips and ends of tools like brush or pencil ), and this is enougth for precision movement capture. I hope that this feature will help to solve the problem with backlash for personal tools by dinamic correction of arm moving in realtime, as well as it allow incredibly improve possibilities of existing industrial robots now. Just need to check this.

[Dan_2013]

New model of 3D printer with precision robobotic arms



http://forums.reprap.org/read.php?178,206458,227054,page=6#msg-227054

[CasperH]

Did you use optical 3D scanners in this solution?

No, everything was basically feed forward for the robot. The delivery parts did have things like optical sensors to check if the part had been removed and would then proceed with offering the next part.

Also, reading your post again I remembered this University robot that was a combination of a rigid gantry system that had arms with some turning points and then possessed the speed possibilities of a delta robot. I have been searching Google now for 15 minutes but I am afraid I cannot find it.

[Dan_2013]

No, everything was basically feed forward for the robot. The delivery parts did have things like optical sensors to check if the part had been removed and would then proceed with offering the next part.

Cool :) I know, this is simple optical sensors, which can determine whether something placed in front of it or not. Now one-cam optical sensor can provide a lot more information, it can even controls the assemblying operations and generate callback for them, and with binocular optic sensor it is possible to create accurate 3D model of operational area in real-time. So the assembling robot can see and control all operations in run-time, and all these operations can be designed in 3D computure simulator first with "exceptions" (like dropped details) handling.

Also, reading your post again I remembered this University robot that was a combination of a rigid gantry system that had arms with some turning points and then possessed the speed possibilities of a delta robot. I have been searching Google now for 15 minutes but I am afraid I cannot find it. :sad:

Thanks, thats not a problem. I've found a couple of educational robots in neighbourhood:



Robin-1





Robin RSS-1 Sphere


As you see first one has gantry construction with increased rigidity, 3DOF, and can be used as CNC milling machine (as it written in description). I.e. this is classic personal FDM 3D-printer. Second has 5DOF, but lower accuracy and can serve as assembling and servicing manipulator.

These are parts of educational FMS but these systems require space about 1200x1500mm on table with CNC milling and lathe machines and materials/parts storage.

I want something smaller :) For plastic and wood first, and than for some soft metals, like aluminium, but metals not obligatory.
The best variant is to find way to use three or more not very strong robotic arms, which can act as separate assembling manipulators, and can stick together around printing head to perform precision operations like 3D printing. This connection will decrease backlash of each robotic hand and increase common strength of printing head, so it may be used for milling or drilling tasks. And if this robot will have same measurements as Trossen PhantomX AX Hexapod Mark II it will be the great and exactly what I want. I'll test Trossen robotic arm first - their actuators has callback sensor, so may be with this sensors and some modifications in arm mechanics I'll able to get needed result, because all backlash in robotic arm now appears in actuators axis offsets under load, but sensors can determine this offsets. So may be this callback data will help to use this hand as micromanipulator with accuracy about 1mm. Need to check.