The Mini Mobile Robotic Printer

[andersendr]

A small portable printer that moves itself, not the paper.  Its like the Roomba but instead of removing things, it add them. 

https://www.kickstarter.com/projects/1686304142/the-mini-mobile-robotic-printer

[electr_peter]

It is printing head ONLY with wheels to move on paper left/right up/down. What kind of printing time and resolution are we talking about? Resolution and movement accuracy are of paramount importance to printers and scanners, that the reason to have sturdy frame. Driving around the paper would result in accumulated errors.
Printing A4 page takes some time. How long do you wait if you want to print 10 pages?

I am not sure why you would want to support such device - it is expensive, slow, would produce lower quality than conventional printing. Who needs low quality personal device-printer to carry with them all day? It would make nice toy though if it does what they promise on KS.

Bonus points for showing semi-working prototype and case.

[andersendr]

It's not my project but it seems interesting.  I work in the field on job sites a lot.  It sure would be nice to whip out something to reprint a drawing that I just red lined.  Or if you need to print out a quick document to sign or you need to have a paper copy of permit.  To me it seems they are in the wrong market as this is an inferior product in the consumer market in time, quality, and price.

According to their info, it takes about 1 minute to print an A4.  The resolution isn't great at 96x192 dpi, but that is more than enough to print B/W stuff.  They clam the final product will have higher resolution, but they don't say.

[ferrix]

It's cute, in a maker/art-project-y way.  But I don't understand what the market is for this.  The only thing I can think of is for really large paper that needs to be very sparsely printed.  But you'd have to have crazy good position tracking to make that work. 

Anything else: have you seen how portable they can get inkjets these days?

[electr_peter]

I work in the field on job sites a lot.  It sure would be nice to whip out something to reprint a drawing that I just red lined.  Or if you need to print out a quick document to sign or you need to have a paper copy of permit.  To me it seems they are in the wrong market as this is an inferior product in the consumer market in time, quality, and price.

I agree that mini printer market is very undefined or niche.
To me, printing on paper signifies that it is "last resort" or "final touch" of something you have finished or are serious about doing. I would try not to print at all unless:
   1) I need to read multiple pages on the screen. That is too much for my eyes. Sometimes paper makes a quick reference that you can write on. Requirements: moderate amount of pages to print, quality could be low.
   2) I need to make official/work document, few pages worth. Requirements: content of page and printing quality must be good.
   3) I am printing some material for other people to read. I would try to use good quality printer, not sub par one.
   4) other cases

Thus, I don't immediately see where lower quality printer could be useful, except for some extra/emergency cases. It is like a very crude 3D printer (slow, low quality, expensive) in a market of good quality and cheap 3D printers(3D market haven't reached this stage yet, but printers certainly did long time ago).

[Duane Degn]

This has received a lot of backers and I'm afraid a lot of people are going to be disappointed in the final product.

The images on the Kickstarter page give the impression the robotic printer can print full page technical drawings and graphics. Yet the video just shows the robot printing single line of text.

Trying to get a robot to travel in a straight line with the precision required to print the type of documents they imply it will be able to print is a near impossibility.

These guys are business students dabbling in robotics and they don't have clue how to accomplish what they are promising. You can see some of the "development" work in posts at Let's Make Robots. The LMR member kkffiirr http://letsmakerobots.com/user/20833 has asked several questions revealing his lack of robotics knowledge.

I don't think they have been honest in describing the risks backers are taking by supporting the project. To make a robot capable of printing as well as they imply their robot will, is next to impossible.

Their "prototype" is a Parallax ink jet printer kit mounted on some omni wheels (normal wheels would have worked for their "demo" video). Here's a video of the printer kit used free hand.



This printer might work reasonably well printing text only documents if you didn't need the print to be high quality, but it's certainly not going to be able to print the type of documents shown in the photos.

I think this qualifies as a scam.

I thought Kickstarter projects had to have a working prototype?

BTW, I like to build robots myself and I've made many different kinds including robots with omni wheels. I won't post links to my stuff by it's easy to find by searching my name.

[CanadianAvenger]

I think this qualifies as a scam.

I thought Kickstarter projects had to have a working prototype?

They do, at about 1:20 in the  video is the prototype printing a message on a piece of paper that reads "hey kickstarter backers!"

It's good enough to qualify as a proof-of-concept prototype. Whether or not they can improve it enough to meet their spec goals is another story altogether.

[Duane Degn]

I disagree. Printing a single line with Parallax Inkjet kit is nothing like making a robot capable of printing a page. Would a 3D printer only able to print in 2D be considered a prototype? They are claiming to make a 2D printer not a 1D printer. A big, big difference. All the work of the prototype is done by the Parallax kit.

If I had a Parallax Inkjet kit, I could reproduce what they've done in a few hours (it would take me a bit longer to make is as small as their demo bot but it still wouldn't be hard). Making a 2D printer capable of reasonable quality print is next to impossible using a wheeled robot as they propose. It might be possible, but they should be honest about the risk of a printer never being produced.

[sleemanj]

Yeah, no way is this going to work, not in any way close to their claims anyway.

1 minute for a full page?  Nope.  1 minute for a line on a page maybe.  It's going to have major problems with uneven surfaces (hope that table is perfectly flat), slippery paper, etc... no way will you get both horizontal and vertical movement accuracy of sufficient quality.  I don't think hard plastic omniwheels are going to give you movement accuracy for a start.

[CanadianAvenger]

Note I was saying it was good enough as a proof-of-concept to pass the "prototype" requirement of kickstarter.  Yes the parallax kit is doing all the printing work. they then bolted that onto a robot carriage to make a printing robot. So it is more than just the parallax kit. Otherwise we could claim that pretty much every FDM 3D printer that came out of the maker movement is "just an Arduino" Don't get me wrong, I don't see what they're pitching as revolutionary, and I'm not even sure they can pull it off to the level of quality needed to get a high quality print, but I don't see any issue in them taking a run at it. As for positional tracking accuracy, I don't see that as an insurmountable problem... an optical mouse sensor is a cheap and very effective high resolution positional sensor that can be used to ensure the bot moves both straight, and the desired distance.

[pickle9000]

Is everyone here too young to know a "turtle" when they see one?

[Jarrod Roberson]

As for positional tracking accuracy, I don't see that as an insurmountable problem... an optical mouse sensor is a cheap and very effective high resolution positional sensor that can be used to ensure the bot moves both straight, and the desired distance.

Sensing relative position, and DOING absolute positioning are two completely different things, the first is easy the second one is a completely impossible task in an open loop system with absolutely no tolerance controls like this one.

High resolution mechanical positioning systems require extreme tolerances and closed loop feedback control, which this toy doesn't have either. I bet that just the drain on the battery would affect the positioning motors enough to cause the thing to introduce error accumulations that would skew over just a few lines of a drawing. Any fluctuations in traction would completely destroy the alignment of a single line much line multiple horizontal lines over a foot of distance, and these would be irrecoverable, the ink would already be down, done deal no way to correct for it
!

One thing is guaranteed, this thing will never print out those line drawings that they have the thing sitting on top of.

If this is as trivial to build and accurate as they claim, someone would have already done it. long time ago, or it would be replicated easily by everyone that wanted one with simple parts and there would be no market for it. Not that there is a market for it now.

[sleemanj]

Is everyone here too young to know a "turtle" when they see one?

Stupid thing is, if somebody made a pen drawing tutlebot now, slapped an arduino on top, and interpreted "logo" commands from an SD card, and threw it on kickstarter, they'd probably sell a ton of the suckers. 

Maybe I shouldn't have given away that brilliant idea.  Ah. I'm, too lazy anyway.

[CanadianAvenger]

Yeah, no way is this going to work, not in any way close to their claims anyway.

1 minute for a full page?  Nope.  1 minute for a line on a page maybe.  It's going to have major problems with uneven surfaces (hope that table is perfectly flat), slippery paper, etc... no way will you get both horizontal and vertical movement accuracy of sufficient quality.  I don't think hard plastic omniwheels are going to give you movement accuracy for a start.

I'll agree on your point on speed, I don't see them reaching the 40sec per page speed mentioned in the FAQ.

I covered positional accuracy in my previous reply.  The prototype has hard plastic [looks like 3d printed] wheels, does not mean the final production unit will. They will indeed have a lot of slip, you can even see it happening in the video. But with a good feedback mechanism from something an optical mouse sensor, the slippage can be compensated for.

As for uneven surfaces, that goes beyond the scope. They are not claiming you can print on any old surface. Obviously the surface needs to be relatively smooth and flat, and free of any debris, otherwise the quality of the print may be affected. This ultimately comes down to the user using the device properly, and not a defect in the device itself. Sorry, can't blame the tool for that one -- but it is a statement on how useful the device really is, as a sufficiently smooth/flat surface to get a decent print, may be harder to come by than they think.

 

[CanadianAvenger]

Sensing relative position, and absolute positioning are two completely different things, the first is easy the second one is a completely impossible task in an open loop system with absolutely no tolerance controls like this one.

Absolute positioning [with a degree of error] is accomplished by the user placing the bot down at the top left corner of the page [note the shape of the bot aids this] Everything is relative from there.  Guess what, your desktop printer does the same thing, it has an optical sensor that detects the to edge of the page, and then assumes relative position from there. If the page happens to feed in crookedly, the print comes out crooked. Just in this case it is more prone to being non-parallel to the edge due to user positioning error.

I do agree that you are unlikely to get high quality prints out of this thing, I would expect something more along the lines of draft quality. [again goes towards the real-world usefulness of it] I imagine that the HP's and Epson's of the world have tried this and concluded that getting high quality output was too variable and too hard for the end user for the product to be deemed as useful, thus never brought to market.

[pickle9000]

Is everyone here too young to know a "turtle" when they see one?

Stupid thing is, if somebody made a pen drawing tutlebot now, slapped an arduino on top, and interpreted "logo" commands from an SD card, and threw it on kickstarter, they'd probably sell a ton of the suckers. 

Maybe I shouldn't have given away that brilliant idea.  Ah. I'm, too lazy anyway.

Yup that would probably sell. For people who want to see a turtle here you go.

http://en.wikipedia.org/wiki/Turtle_%28robot%29

As for making a small printer try making an old fashioned plotter. It's completely possible to shrink a plotter / vinyl cutter and keep reasonable output. No you can't print a photo, you can draw a picture.

Amazingly cheap on ebay for a big one.

http://www.ebay.ca/itm/New-24-Redsail-Vinyl-cutting-plotter-Best-Value-Sign-vinyl-cutter-Kit-Artcut-/251454461377?pt=LH_DefaultDomain_2&hash=item3a8bda95c1

[Jarrod Roberson]

the "degree of error" will accumulate so quickly at 192dpi that it will be crap in just a few mm.

I ran high resolution single pixel laser scanners back in the day and can tell you that doing absolute positioning at the resolutions that they are implying the marketing materials is impossible from vibration of the motors alone, the variability of the drive train mechanical slop and the electrical variances in driving the motors. Even at the extremely slow axial speeds those laser scanners ran at and the extreme tolerances, I still had to rescan stuff all the time because the table or machine got bumped and there would be gaps. Same applies here, just in reverse. I am pretty confident that just typing on the laptop on a shaky cafe table would throw this thing off over 1 - 2 mm of distance.

This thing would need CRAZY accurate mechanicals that would cost 1000X what they are aiming for, and probably would just barely meet their target specifications and the implied quality of the technical drawings they are implying are possible. And would be 1000X slower than they can imagine.

And that would be for a single line, as soon as they moved down the page, all bets are off, there is no way to position a open loop system like that at that resolution, the mechanics are just too sloppy, there is no way to sense or measure external inputs.

There is a reason inkjet printers run on isolated rails and only in one axis and the paper feed mechanisms are prone to leave gaps even in modern inkjet printers. And guess what, once that ink is on the page that gap is there there is no "compensating" for it, that bell can't be unrung, you just wasted expensive ink and more expensive time.

[PlainName]

no way to position a open loop system like that

As someone pointed out above, perhaps it isn't an open loop system. Using, for instance, a mouse sensor for tracking makes it closed loop, doesn't it?

[CanadianAvenger]

It looks like they are indeed using an optical mouse sensor for feedback... this was confirmed earlier today in the comments. [about 9 hours ago]

[FreddyVictor]

as others have already said, interesting idea, but probably going to dissapoint

the 'test' print in the video was done in ideal conditions ie perfectly flat table etc and I would imagine it would have issues if the surface wasn't level

[Jarrod Roberson]

no way to position a open loop system like that

As someone pointed out above, perhaps it isn't an open loop system. Using, for instance, a mouse sensor for tracking makes it closed loop, doesn't it?

No it doesn't, because the accuracy and latency of the positioning mechanics isn't accurate enough to respond at the resolution of a mouse sensor. Just the variability in the slop of the drive train is going to be unaccountable for with that low of resolution/tolerance parts. The slightest input onto the table surface is going to throw this thing off more than it can account for quicker than it can recover from.

This is just a big FAIL.

[PlainName]

No it doesn't, because the accuracy and latency of the positioning mechanics isn't accurate enough to respond at the resolution of a mouse sensor.

Forgive me for being a bit dense here, but are you saying that the slop in the drive chain means that there isn't a closed loop? Would it be closed if the mouse sensor was sufficiently inaccurate to match the drive chain slop?

This question isn't about whether the product can perform to spec or whether they will rely on pre-packed magic. It is merely asking if some method of feedback coupled with some method of control taking into account that feedback constitutes a feedback loop (even an awfully shit and not fit for purpose one).

[PlainName]

The slightest input onto the table surface is going to throw this thing off more than it can account for quicker than it can recover from.

This is just a big FAIL.

I fear that you have created for yourself a strawman (or, if you prefer, a red herring; a MacGuffin). I recommend that you take a chill pill before reading on 

It doesn't matter that there is slop in the wheels. They don't even need very accurate control of them. What they do need to do is ask one question and execute three actions (though two of those are implied by the question):

Code: [Select]

Q1: Do I need to drop a blob of ink here?
If yes:
   A1a: Drop blob of ink
   A1b: Tick off location from list of locations to be blobbed

A2: Steer towards next blob location

Et voila. The mouse sensor tells them absolutely where they are (at least, absolutely relative to some reference, maybe the top left corner of the paper). As you note, it is very accurate. All they need to do is wave that bot in roughly the right location and wait until the exact point passes underneath, whereupon it gets blobbed. Think of those LED wands which display messages when you wave them in the air, but instead of light they're using ink:



Worst case, they spend a fair amount of time trying to get their sloppy wheels to get over the spot (but, they'll have a list of blob spots so can drop other ones on the way to the one they're aiming at, to save time). But don't forget they are using an inket head, which has a width and a number of jets across that width. If the inaccuracy of the drive train is still within the width of the print head then they don't even need to jiggle because they can just select the right jet as it passes over the blob location.

Blimey. Wish I'd thought of this first.

[Edit: fixed tyops]

[scientist]

I could imagine using it for PCBs if it was a laser printer and not a cheap inkjet. 

[PlainName]

if it was a laser printer and not a cheap inkjet

Get a laser printer into that package and I suspect you could make a fortune without needing to have it trundle over the paper at all