Toy Swap? Have: Rovio, Want: World Domination (or a plotter frame)

Community Bar Sport
1

So, as you folks know, I picked up a Wowwee Rovio Mobile Webcam (basically a WiFi tank with a webcam) a few months back. Though missing it's battery pack, it's otherwise fully functional, I had it publicly usable for a time running on a home-made battery pack made of AA NiMH cells. Cute, fun... but in all honesty, I keep on looking at it wishing I had something a little more useful to my interests rather than the ultimate radio-controlled car.

I also am looking for a way to carry out something I've envisioned for some time.. either an x/y table or (even better) a robotic arm (or some other reasonably usable positioning system). What I want to do is build and teach a robotic system to paint. In acrylics. Not nozzles, not jets. Brushwork. Initially I'll have to mix the colors, but in time, I'd automate as much as possible. I want to be able to handle at least 8x10.. a plotter MIGHT suffice if it had high clearance and was, say, 11x17". An arm of course gives all the range I need.. but a bit pricey compared to a mobile webcam, methinks..

Conceptually, it's really not that bad.. I'd actually likely use watercolors first, simply because doing the work via Washes is probably completely possible.. color intensity is deposited pigment (so can be wash-layered), and monochrome would be possible too, so a "first version" would be reasonably easy assuming a motion control system with even 1/4" accuracy. It'd have to start as flood/dot to work out the kinks of motion, but I'm betting vector-based "Brushstrokes" can be worked out. Sure, There's better ways to print. I want to do it cooler, not better. I want to make an art printer that in and of itself is a work of technological art.

Carried to where I'd REALLY love to take it.. envision this system (a camera, hidden computer, and robotic arm) set up with a bit of window dressing to appear more like a human automaton or even a Terminator-esque Cyborg.. the surreal vision of an artificial artist painting in the styles of the Masters. A few blinky lights and chrome. A stainless-steel skull wearing a jaunty beret, red eyes (diffused LED) staring blindly nowhere while the arm (clad in a smock as much as possible, but NOT completely) creates a recognizable knockoff of Water Lilies or the Mona Lisa... or your own portrait, live. Stunning, creepy, and you wouldn't be able to look away. It's meant to be art in it's own way.

What I'm wondering is if there's any tinkerers out there that might be interested in swapping something I could use in terms of x/y (I'm planning hobby servo mounted brush) for that little Rovio... someone should be having fun with the Rovio, and I'd prefer something I can apply towards this slightly demented "Art" project.

I'm figuring that a plotter table might work, since an arm is a little crazy, cost wise. To make the "artist", I'd run the empty sleeve of the painter's smock and just attach it to the "print head", so the "arm and hand" would move with the brush, rather than vice versa, but it'd still work visually. Bascally prop up the plotter table on an easel, then try to creatively cover as much of the mechanism from view as possible..

An xy table is going to run on steppers, Arduino to the rescue. Yes, it's a crazy idea, but I can't seem to shake it. In terms of managing the physical automation system, Arduino is going to be the right answer, I'm betting.. the code will be.. interesting. Accounting for amount of paint on the brush by estimation will be needed, the trial and error alone on that ought to be enough to drive me the rest of the way over the edge..

2

I don't have anything to swap you (certainly not a plotter table), but you might think about giving the arm doing the actual "painting" (rather than a plotter moving the arm) a real go. I admit it won't be easy, but for a start, imagine that the paint "brush" was simply attached to the end of a pan/tilt mechanism, ok? Now, if the canvas was spherical in shape, then the "arm/pointer" could reach any coordinate on the inside surface of it, right? So - in order to reach all the coordinates of a flat canvas, you would have to add some mechanics and such to move the arm/pointer/brush closer/further away as it was directed (likely some form of spherical/cartesian coordinate transform of some sort). Just an idea, though your original idea has merit, too (maybe you could direct the arm by using the x/y plotter mechanism to move a "magnet" under the canvas, and the "hand" holding the "brush" would have a carriage or something to allow it to move across the canvas freely, while still not smearing or such the image it was drawing. At the end of the drawing, the magnet could release the hand, allowing it to drop swinging away from the canvas. An operator would have to replace it to start it again for the next piece)...

3

The arm of course is the most elegant solution.. but I think you got it backwards. If the arm is controlled from a central pivot (or any pivot for that matter) then the reach at any given point is a curve - the radius between the "hand" and the "shoulder". I've assumed that something like that needs to worked in polar coordinates and then you'd have to essentially work out the math for a flat "projection" as a plane intersecting a spherical potential work area. Any given x and y location on the canvas would have a z component which is the radius compensation for the canvas' flatness. We as humans work like this- pivoting shoulder, elbow, wrist, etc. A plotter device "lives" in FlatLand essentially, it's working area is a plane, while an arm's working area is a sphere. To paint a spot at x,y on the Plane, the sphere of the working area must intersect the plane to at least that degree.

A plotter would be the opposite. X and Y don't need to be calculated (surface projection) and there is no Z component, simply "pen up" and "pen down". My mannequin or robot artist or whatever.. the arm of the Artist wouldn't guide the pencil.. it would follow it. Basically an empty smock sleeve that just gets carried along with the "plotter" head's movements. The dummy "arm" is purely for effect and is only a prop.. the actual working parts are just a good old plotter mechanism. Remember also, just like a real painter, the effect will be to have the canvas/easel/plotter mounted like an easel, with the Artist nearly parallel to the canvas rather than 90 degrees off, as a tabletop plotter would normally be. A plotter setup would be the most accurate painter by far.. but super resolution isn't really a goal. Go ahead and look at a typical painting, and tell me what the effective "resolution" of the image is. I'm hoping for 1/8" resolution at a minimum- giving an 8x10 canvas an effective resolution of 64x80 "pixels". Even taken to the level of the Masters, you're talking about twice that, or 128x160. Yes, it's a LOT more nuanced than that, but in terms of raw resolution-- painting is pretty low tech. Step one is getting at least that accurate, then the next headaches start..

To give you an idea of when I first started thinking about this, I will say that it originally occurred to me that it would be really nice if you could use a nice quiet paintbrush for printing rather than the danged LA75. As an aside, I'm getting more and more keen on the idea of hitting some of the salvage shops and getting some of that old DEC hardware for Arduino use. I so want an original Cakebox/Maytag 10 Megabyte "Winchester" drive or a DLT drive for Arduino data storage. 5 1/4" floppy would have some coolness factor, but you'll have to go 8" or (God help me) 12" floppy in Xerox Sigma format for a gold star. I'd love to utilize some old rack CORE memory- just because. I had a chance to pick up a punch/reader a few years ago and kick myself for not doing it. Big blue and the rest just never had the full on "style" that DEC had. The kids here will never know what those days were, the huge computer companies of the seventies. Now, I see the carcasses of these industry giants as junkpiles. Quite the ride, huh?

Arm is FAR more elegant, but I'm thinking cost will be crazy-- and as you noted, coordinate conversion alone could present a whole bucket of fun...

I've even considered the whole beast could be made inside a frame (pvc, steel, whatever) and The Artist's movements could be done by flywires.. three (or four, depending on the design) cables (fishing line) in tension, like is used for Stadium Flycam setups. I figure the problems in blocked angles, bounce in the cables, and the rest would kill the design before it could be made to work. An arm would be the coolest way to do it, but the plotter version is probably the most likely to succeed in working, I think..

4

focalist:
The arm of course is the most elegant solution.. but I think you got it backwards. If the arm is controlled from a central pivot (or any pivot for that matter) then the reach at any given point is a curve - the radius between the "hand" and the "shoulder". I've assumed that something like that needs to worked in polar coordinates and then you'd have to essentially work out the math for a flat "projection" as a plane intersecting a spherical potential work area. Any given x and y location on the canvas would have a z component which is the radius compensation for the canvas' flatness. We as humans work like this- pivoting shoulder, elbow, wrist, etc. A plotter device "lives" in FlatLand essentially, it's working area is a plane, while an arm's working area is a sphere. To paint a spot at x,y on the Plane, the sphere of the working area must intersect the plane to at least that degree.

No - I understood what you were trying to do; I was just trying to present the idea of a real arm that might need only a few degrees of freedom (a 2-axis shoulder, and maybe a elbow), along with some simple linkages to keep the brush/pencil on-canvas. Your idea would certainly work, but I am not sure about the whole thing looking like a robot when it would be apparent it was a plotter; an arm device would avoid this.

focalist:
A plotter would be the opposite. X and Y don't need to be calculated (surface projection) and there is no Z component, simply "pen up" and "pen down". My mannequin or robot artist or whatever.. the arm of the Artist wouldn't guide the pencil.. it would follow it. Basically an empty smock sleeve that just gets carried along with the "plotter" head's movements. The dummy "arm" is purely for effect and is only a prop.. the actual working parts are just a good old plotter mechanism. Remember also, just like a real painter, the effect will be to have the canvas/easel/plotter mounted like an easel, with the Artist nearly parallel to the canvas rather than 90 degrees off, as a tabletop plotter would normally be. A plotter setup would be the most accurate painter by far.. but super resolution isn't really a goal. Go ahead and look at a typical painting, and tell me what the effective "resolution" of the image is. I'm hoping for 1/8" resolution at a minimum- giving an 8x10 canvas an effective resolution of 64x80 "pixels". Even taken to the level of the Masters, you're talking about twice that, or 128x160. Yes, it's a LOT more nuanced than that, but in terms of raw resolution-- painting is pretty low tech. Step one is getting at least that accurate, then the next headaches start...

Like I said, I know what you were going for; it certainly avoids any IK calcs, which could be a bear to understand and implement. Its only drawback is that it would look like a plotter guiding an empty sleeve - which is why I was suggesting the magnetic device, with the x/y control on the backside of the canvas, hidden from view.

focalist:
To give you an idea of when I first started thinking about this, I will say that it originally occurred to me that it would be really nice if you could use a nice quiet paintbrush for printing rather than the danged LA75. As an aside, I'm getting more and more keen on the idea of hitting some of the salvage shops and getting some of that old DEC hardware for Arduino use. I so want an original Cakebox/Maytag 10 Megabyte "Winchester" drive or a DLT drive for Arduino data storage. 5 1/4" floppy would have some coolness factor, but you'll have to go 8" or (God help me) 12" floppy in Xerox Sigma format for a gold star. I'd love to utilize some old rack CORE memory- just because. I had a chance to pick up a punch/reader a few years ago and kick myself for not doing it. Big blue and the rest just never had the full on "style" that DEC had. The kids here will never know what those days were, the huge computer companies of the seventies. Now, I see the carcasses of these industry giants as junkpiles. Quite the ride, huh?

Good luck finding it; most of that stuff is long gone, and I would think what little remains might command higher prices (only talking about the bigger stuff - mainly because of the metals it contains); also, finding cheap core memory is getting more difficult. I have a board of it (16K, IIRC) hanging on my dining room wall - and an old block of large core memory (bakelite frame) in my shop as part of my collection of old computer junk. I got both fairly cheap quite a few years ago; nowadays I could probably make 10 times what I paid for them.

focalist:
Arm is FAR more elegant, but I'm thinking cost will be crazy-- and as you noted, coordinate conversion alone could present a whole bucket of fun...

I don't think it would have to be too expensive - I think a reasonably simple arm might be made for under $200.00; the trick would be to figure out how to create an arm that moved somewhat naturally, but mechanically "solves" the IK issue to keep the hand on the paper as it moves in the x/y plane (essentially it would have to mechanically - or with a minimum of a third "extensor" elbow joint - extend the hand in/out to keep it the same distance to the canvas as the shoulder/upper-arm moved in the polar/spherical fashion).

focalist:
I've even considered the whole beast could be made inside a frame (pvc, steel, whatever) and The Artist's movements could be done by flywires.. three (or four, depending on the design) cables (fishing line) in tension, like is used for Stadium Flycam setups. I figure the problems in blocked angles, bounce in the cables, and the rest would kill the design before it could be made to work. An arm would be the coolest way to do it, but the plotter version is probably the most likely to succeed in working, I think..

Now, something like that might be really interesting - basically an Arduino-controlled "life-sized" marionette; from an art perspective, you could even explore who-controls-who issues, autonomy, etc.