Mindstorms/Technic: Lego factory project

[Dryw Filtiarn]

I've been working on a new project for a while already, and after a lot of thinking, testing, building, rebuilding and even more testing, building and some more rebuilding, I'm now ready to start showing a bit of the project to the public. Unfortunalty I can't post the visuals on this forum as it's a video demonstrating what I have finished at this point, so here's the link:

Short explanation of the project:

The goal of this project is to build a lego factory that differently from the ones you can see here:

• 
  
• 
  

Has the following properties:

• Compact, this means that I'm building a more compact factory then the given examples of projects done by others. In my oppinion those factories are despite their perfect functionality to large to what would be absolutely needed. When you look at those factories, you will note a lot of repetation of identical functional units (i.e. multiple press-towers). My factory is aimed a building a general system which uses basicly a single building unit that can handle nearly every part as long as it has at least 1 stud on the top of the element.
  
• Beauty, as previously said projects shown earlier by others are bulky and rather large. Next to that they are purely about function above design, which doesn't make it too attractive to look at. Earlier factory projects also tend to be build in rainbow coloring, which doesn't really help either. My project is aimed not just on function but also on elegance. This means that a large part of the factory is actually purely aimed at functionality, but next to that I also aim at the visual aspect, by making the entire project look like a large factory building, furnished and decorated to minifigure scale. This means that it will contain control-rooms, etc...
  
• Variable, factories previously shown, only have a single function, being able to build a single car of airplane (at least in the given examples in this post), the only variaty provides is the option to pick colors. My aim is to build a factory that can build more then a single model. Eventually it might be able to build a small house, a car, a plane, a boat and maybe even more.
  
• Detail, beside the detail already mention in the "beauty" item, I'm also focussing on detail in the models the factory will be able to build. Previous factory projects produce rather primitive models. A car consisting of 2 sets of wheels, a plate, 2 slopes and a brick. This is not what I want to be able to do. I want more! Goal for me is to be able to build cars that would match the tiny-turbo series Lego produced. This means that a single model is this factory might consist not of just 6 pieces, but that a model might contain 50 parts or maybe even more. (At this point I would be able to build models (of any type) which can be 14 x 14 studs wide and up to 3 2/3 bricks in height)
  

[furious]

looks impressive so far.

[DB24]

Looks nice!

Great idea to make a machine so universal, after all, isn't that what Lego is about?

Is it fully automated in the video? Or are you controlling it in any way?

I look forward to seeing a more completed version soon

[Dryw Filtiarn]

In the video I'm controlling it manually with 2 PF controllers. The next step will be to automate it using an RCX.

[CP5670]

This sounds very interesting. What kinds of pieces will the cars use? If you're using more than just basic bricks and plates, I think one of the main challenges would be finding a way to reliably pick up and position pieces of different shapes with the same claw arm.

I've seen those other "factory" machines on Youtube and although they're extremely cool, as you said they seem too large and bulky considering what they do.

[Dryw Filtiarn]

The system I am building will be able to place every element that has a stud on top. These can be bricks, plates, slopes, axle-plates, and so on. Positioning the elements isn't really much of a problem to be fair. Small elements like a 1x2 plate or a 1x1 brick, can perfectly be placed with a single action. In some case with larger elements, i.e. placing a 1x6 plate, it might require placing the element in the position first and then move along a couple of the studs of the element to press in place everywhere. I've done some manual tests with the system yesterday and it seemed to work pretty wel. At this moment I'm just about to begin programming the RCX that will be attached to the trolley that carries the building arm back and forth.

Edited May 6, 2009 by Dryw Filtiarn

[green dewback]

I watched the plane factory video and it's very impressive.

[Dryw Filtiarn]

I agree it's impressive (note that it's NOT mine!). Though I think it can be done better and more efficient.

[jimjxr]

the aircraft/car factory is not that impressive, sure they're big, but the final product is so small and simple, it's a big let down. I think we can do much better than this.

[green dewback]

But its automatic !

We can't expect it to build much better than a toddler as robots aren't as nimble as our fingers.

Edited May 8, 2009 by green dewback

[Dryw Filtiarn]

I fear I don't agree on your statement that a robot can't build any better. Real manufacturing robots can certainly do better of course and even with Lego if you would create a somewhat more clever concept it's perfectly doable to build rather complex constructions.

Ofcourse you can't expect a Lego robot system to build complex SNOT models, but there is a lot that can be done with it, as long as you use a clever design for your robot.

With a proper design most of a model like this should be perfectly posible to build by use of a robot:

And considering what I have build in my factory project and the results of early tests, I have already proven that it really is possible to do this.

Anyhow, current project status is that I'm about halfway programming the RCX with a testing program that controls the trolley with the building arm attached to it as well as the building arm itself. As soon as I have finalized this, I will give it a go to see how it works. If it's satisfying I will make a video out of it and post it.

[forbid89]

This is a really rad idea, I saw those videos a couple years ago and wanted to develop my own factory but never got around to it. I think what you're doing is a better way. Those massive factories are designed to make just one design but by using a more universal design the output possibilities are endless. I don't know how feasible it is but I think developing something similar to an industrial robot would be the best way to accomplish the task. As with any engineering task you're only limited by imagination!

[Dryw Filtiarn]

This evening I've been working on a new design for the track that carries the trolley. After testing yesterday it should that the version I had, wasn't reliable enough once all the weight was on. The old design was based on a single beam track with the trolley hanging from it with U shaped braces (covering top, one side and bottom), but having the weight on it it turned out that the tracks were being pulled in causing extra free space between the track and the braces. This design also had no options to be strengthened as that would only be possible on the inside of the track, were some parts of the building arm are.

The new concept design I've made (I only created a small section of it for testing) uses a U shaped rail that is open on the inside, whereas the trolley only has a single beam that slides in between. This design has the advantage that once there is weight on it, the tracks push out (which can easily be braced). Next to that I it will be easier to add the sensor triggers for the touchsensor inside of the track so that they can be easily hidden in the design. And to me a major advantage of this new track next to the fact that is more stable is also that it can be properly build in a modular fasion, which will help in taking the factory to events.

All in all I'm quite satisfied with the result, though it will take me a good few hours to build the new track for the full length of the model as well as installing in the structure, but it's well worth it.

When I finished the track I will retest the whole thing and if I'm satisfied with it, I'll make some photo's and video's of it and will show them to you.

[Dryw Filtiarn]

Time for me to give this topic a boot up it's ***. Unfortunately since the last status update on this factory idea no results came out and I have abandoned the project.

The good news currently is that I have started working on a new factory again a few months ago and I'm currently working hard to get it finished as it is to be revealed/displayed at LegoWorld (Zwolle, NL) in October!

At this time I have no recent photo's or video's of the progress I'm up to at this point, but these will be made soon again when I get a couple of elements of the factory model completed to satisfaction. What I do have though is a couple of video's from early testing, one made at home when I was giving the factory a trial run, and the other one as it has been displayed on a small local Lego event in August.

It's first trial run at home:

How it functioned at the August event.

A note to the video's... The video's are showing a grabber attached with which it handles the boxes on the sliding floor element. This grabber was just temporary for the August event so I could do some event-condition testing (events always tend to mess up complex technical constructions due to warmth and moisture) and had at least something to display. Immediately after the event the grabber was replaced with a building head which is quite comparable to the original version in my previous attempt to build a factory:

Some statistics about this project:

* 5 NXT's controlling 13 motors

* 2 meters in length

* 40 centimetres wide

* 26 centimetres high (not counting in the booms of the robots)

* around 25.000 parts (rough estimate)

* handles 103 different parts (11 part types * 8 colors + 3 part types in 5 transparent colors)

* stores a total of 1648 parts in it's warehouse section

More information, photo's and video's will come soon...

Hopefully by the end of the week my orders will arrive so I can complete some elements of the factory this weekend... Waiting for over 2000 parts to come in...

Edited September 14, 2010 by Dryw Filtiarn

[Dryw Filtiarn]

Here is a short update, while I'm still waiting for the parts to arrive, hopefully they'll come in tomorrow...

Anyway, in the meanwhile I have been working on completing the middle robot that goes in between the warehouse and the actual robot that will be building the models. This one will take care of receiving the parts from the warehouse and aligning them so that the building-robot can pick them up and place them in the model. This robot will also take care of rotating parts so that parts can be used in the required orientation.

More updates will come as I further complete parts of the factory, which will likely be near the end of this week.

[DLuders]

@ Dryw Filtiarn: Wow, this is getting more intricate and refined all the time! Keep those updates coming! You'll have to keep a big supply of batteries around for the BrickWorld 2010 event in Zwolle, NL, since kids will be wanting to see it running all day long!

Just out of curiosity, do you work in a factory? Some of the machinery that you've made is quite similar to what could be found in modern industrial processes....

[Dryw Filtiarn]

Batteries won't be an issue, all the NXT's will be running on the rechargeable batteries and permanently connected to an AC socket. But indeed the idea is that the factory will be running all day long. That is, it will probably run in 2 sessions, one in the morning and one after noon, both lasting for a couple of hours. Also to give myself some relaxation during the event and time to eat ;)

In regards to whether I work in an factory, nope, I'm a web front-end developer at an IT company. But I have worked in several production plants so some of the inspiration comes from there indeed.

[Dryw Filtiarn]

A short update to the project. I've managed to get 3 of the NXT's programmed now, which will now allow me to retrieve the parts from the middle robot (seen in my last video update) and place these parts in the model. Currently I have some minor issues with the calibration of the building robot, so in 99% of the tries the building robot currently fails to place the part correctly in the model, but I expect this to be solved later today.

Once I get everything working reliably I will make a new video showcasing the building process.

As soon as I get the building part of the factory fully functional I will program the last robot, which is the one in the warehouse. That robot is relatively easy compared to the other two, so it shouldn't take as much time to get them programmed, although I will have a challenge there to get the 5th NXT in the communication setup.

This is how the NXT's work together:

        =========================================================
        | Computer (sending building instructions step by step) |
        =========================================================        
                                      |
=============================================================================
| Master NXT (also responsible for build platform & Takeover robot rotator) |
=============================================================================
             |                          |                  |
==================================       |                  |
| Warehouse robot (Front section)|       |                  |
==================================       |                  |
             |                          |                  |
==================================       |                  |
| Warehouse robot (Rear section) |       |                  |
==================================       |                  |
                                        |                  |
                   ==============================          |
                   | Takeover robot             |          |
                   ==============================          |
                                                           |
                                    ==============================
                                    | Build robot                |
                                    ==============================

[DLuders]

Your long-term dedication to your factory project shows great determination -- I'm sure that the final result will be magnificent! Keep us posted!

[VASH321]

I really like your project, I'm starting something similar although it wont handle 103 different parts ...

One question though, how do you store the parts, on pallets? in a high bay storage?

anyway great project and Good Luck!

[Dryw Filtiarn]

I really like your project, I'm starting something similar although it wont handle 103 different parts ...

One question though, how do you store the parts, on pallets? in a high bay storage?

anyway great project and Good Luck!

The storage indeed is something like a high-bay storage. The warehouse consists of 45 base-plates with grooves in them which hold the parts in rows of 16. The entire warehouse is 5 base-plates stacked and 9 next to each other. A video can be expected of it as soon as it works, then things will start to make sense.

To update on the status, I left the building-robot programming for a bit now, in order to do some experiments for the warehouse robot. I'm attempting to get a communication going between the two NXT's involved in that robot through the sensor port 4 communication (RS485 for the people who know what I'm talking about), that way I can avoid getting a too complex web of NXT's communicating over blue-tooth where one NXT will be both a slave to the main controller as well as being master to it's supporting NXT. If I manage to get the wired-communication going by use of the sensor port that would mean I would only need 4 NXT's communicating through BT.

This would be the setup:

Controlling computer <--Blue tooth--> NXT1 (Master controller)

NXT1 (Master controller) <--Blue tooth--> NXT2 (Takeover robot)

NXT1 (Master controller) <--Blue tooth--> NXT3 (Build robot)

NXT1 (Master controller) <--Blue tooth--> NXT4 (Warehouse robot, master NXT, front side of the robot)

NXT4 (Warehouse robot) <--Wired sensor 4--> NXT5 (Warehouse robot, slave NXT, rear side of the robot)

The first 4 blue tooth connections are up and running, now I'm hoping to get the final communication to the last NXT working as well.

[Zerobricks]

The complexity is just mind-boggling! Which you think is gonna be harder mechanics, or programming?

[Dryw Filtiarn]

I tend to say that both are equally complex. The programming itself isn't much of a deal, neither is the building of the construction itself. The complexity is mostly to be found in the combination of the two, because the software and the hardware (the construction) need to work together in order to achieve the level of accuracy needed to be able to achieve the goal.

So if the robots are not capable of fine movement (mechanically) you won't manage to realise this, and if the software isn't capable of handling the fine movement (programmatically), you can't realise this either.

[Dryw Filtiarn]

A little update: I just finished the RS485 communication (sensor port 4 <-> sensor port 4) between the two NXT's in the warehouse robot. I though I had solved it yesterday already, but for an odd reason when I turned of my NXT's and started them again, the whole communication failed again. This time I'm sure I've got it functioning properly. I'm now moving over to the building robot again, to finish the programming on that, which is mostly focussing on placement accuracy, which might turn out to be trick to get solved again, but I'll see. More info later.

Edited September 27, 2010 by Dryw Filtiarn

[Dryw Filtiarn]

A new video update regarding my factory project. Unfortunately the quality of the video isn't great, due to bad light conditions, but a better one will come as soon as I improve the robot displayed in this video.

This video shows you the building-robot at work, placing bricks (on a fixed location in this video) on the building platform. There's still some work to be done in order to get the release mechanism inside of the building-robot's head to function properly as it is currently either breaking apart the bricks that it build or lifts the building platform. It's mostly a matter of finetuning luckily. A new video will come later showing some more advanced building.