Project: Car laser scanner.

[Conceptual]

I wish there was a project forum on here, I think that it would be great to have a place to actually BUILD projects while getting real time input from the community here.

Anyways,

A friend of mine has a garage with a paint booth. Now, he doesn't do collision repair any longer, and doesn't use it at all. I was talking to him about building something inside there, and he asked what my idea was, and he told me to look into it.

I was thinking about building a machine that could laser scan an auto, and then have multiple spray heads that would either paint the car, or later on, apply OLEDS.

I know that it would have to be a hardware and software system, but I am looking for some ideas from the community here.

Questions:

1. Is this even feasible for under $20k?
2. How complex does a workable system need to be?
3. Software currently available that can operate this system?
4. What components needed for the scan system? The spray system?

If anyone is interested in contributing ideas or experience, I would be greatly appreciative!

Thank you!

Chris

[Jersey Tom]

It's a one off thing you're building...

If you're not going to be doing any of the fabrication (welding and machining) yourself, or programming...

I would expect total cost to be well upwards of $100,000. Maybe in the 1/4 million range. Not sure.

CNC machine tools in general run for maybe $40,000 minimum, and that's a production thing. Design and programming already done, tooled for making multiples. This is going to have a much more intricate control and servo mechanism (a 6-axis deal basically).. plus it would be very difficult to program software to intelligently recognize vehicle geometry and know when and where and how to spray. Massive programming and development cost, plus tuning it. I'm not even sure if auto manufacturers use the laser scanning thing.. rather than just have an actual painter do a first pass that's then recorded and digitized for recreating.

If you had a bunch of buddies that were cool non-linear control system programmers, image/shape recognition guys, design engineers, fabricators, etc.. then maybe you could do it for < $40k but that's pushin it.

To put it mildly.. it would be a very difficult and challenging undertaking.

[Carlos]

That's a pretty wide open question. I found this scanner and software at $2500USD:
http://www.ministryoftech.com/2006/12/2 ... r-scanner/

A Roland scanner and software for $22000 USD.
http://www.studica.com/products/product ... ctid=10288

Try the search terms ' laser scanner CAD integration ' and ' laser scanner design integration ' there is a lot of material available. Just sift the sources.

This company may answer your questions, they build laser scanner paint robots and similar production machinery.
http://www.integratedindia.com/sitemap.htm

[Conceptual]

The idea came from us talking about a paint effect that he always wished he could have on this car. That is "moving flames". I told thim that if it was possible to spray the OLEDS onto the car, not only could the flames move, but they could actually be "acted" upon by the speed and acceleration of the car.

I do have several engineering friends that are grads from Penn State University, but all of them are so tied up with grad school stuff, that it is very tough just to get a few hours with them.

I didn't think that the actual boom would be that tough, to be honest. When i was little, my family built a similar contraption for farm equipment, but that carried sand-blasting nozzles instead of paint heads. It was rather simple in design and operation, and there was no computer control, but they built it in an afternoon, and I remember much of the design. I was hoping to be able to mount laser scanners, and then use that data to apply paint/OLEDS.

I appreciate the links too Carlos. Thanks for the input so far.

Anyone else care to contribute?

Chris

[Jersey Tom]

Making a swinging arm thing isn't too bad. The computer control is the trick

Especially when you're doing servo control and really trying to minimize mechanical backlash so your control system can actually deal with the thing. It's got to be tight.

[bettonracing]

I know u're not trying to get this technical but Faro Technologies (http://www.faro.com)already makes the measuring laser side of what You're trying to make. I'm not sure if any of their customers are using it for paint but quality assurance and robot assembly are the two most frequent automotive uses of their laser scanners.

You may want to look into automated car washes. Effectively they're trying to achieve the same effect as the auto-paint (i.e. evenly covering all relevant surfaces), except with soap, water and wax. The research I did pointed at $100K+ for the carwash... And that was using commonly available parts and containers and doing most of the fabrication myself. Adding in the controlled paint spraying (and possibly the feedback or additional systems to prohibit overspray) could drive the cost up dramatically. Look up what the auto manufacturers are doing and try to bring down the cost.

You might also consider a machine that replicates the hand motion of the painter, although still human controlled.

Regards,

Kurt

EDIT:That's what happens when one takes too long to post. People come in and say parts of what u were typing.

[Ciro Pabón]

I know this is not what you want to build, but what if you dispose of the scanner and replace it with a CAD drawing of the car?

The machine wouldn't be universal, I mean you have to have the shape of the car already into the computer, but then you could make it for less money: all you need is for a guy with a scanner to provide you with the coordinates for the most common models, or you could get the shape of the car (the CAD drawing) from other sources, including image processing.

Anyway, if you buy the scanner you would have to provide the computer with the information about where are the areas you do not want to paint (lights, windows, bumpers, etc.): if you don't you would be wasting a lot of paint on paper covered areas. I imagine that if this is the case, the software should include some way to make the scan and allow you to indicate to the computer where these parts are. In the end, I imagine it would be more intelligent to have pre-programmed layouts.

Another thing I cannot avoid thinking about is that when Polaroid developed the SX-70 instant camera back in the 70's, they used a sonic transducer and a microphone, to focus the camera. The kit was sold by Polaroid for hobbyists and it costed only a few tens of dollars. It was used on many more Polaroid models and it is still popular with robot hobbyists. You can even get it from an old camera. There is a lot of literature on how to use that kit to control a robot, for example, here: http://www.aes.org/e-lib/browse.cfm?elib=3680 It has an accuracy of 1% over the entire range of distance, or so I've read.



To sum up, an alternative would be to use preprogrammed layouts and a sonic transducer to keep the painting head at a more or less constant distance.

[Conceptual]

Thanks for the replies. Looks like this might be way more expensive than my friend is willing to invest. He wanted to do some of his "show" cars with the OLEDS, but to incorporate that on top of the rest would make it a $300k paintjob!

Thanks for the replies!

Chris