robot skin

how would one put that over a hand, fingers, etc? it looks like every separate skin piece should be irregulary shaped.

okay, but what if I need some piece of the shape like this:

*** ******** ************* ************* ************* ** ************ ***** ********** ***** ******** ** ***** ?

sounds good, but it wouldn't readings also change on simple motion like crooking a finger, etc?

Dennis, Sounds very interesting. Do you have anything written up on it that is viewable on the Web? I'd think a lot of people will be very keen to try it out. You might become famous!

-- Gordon

Dennis wrote:

Hi Gordon,

I don't have it on the web yet. Just in a notebook and assembler code on a computer. I have a web site, presently used for a small business I started several years ago. I have recently licensed the technology I was using for my company to a large company that will be mass producing and mass distributing my present products, starting in January. I plan to change my web site into a robotics site and will place the skin on it. I'm planning on using Visual basic to be able to map the pressure points on a bitmap image with red for higher pressure points and blue for lower pressure points. Presently the PIC16F877 is sending the data through an rs232 port in ASCII format. I'm planning on changing to a USB port also, probably with a PIC18F2455. Speaking of fame, I have your book Robot Builders Bonanza and love it.

Thanks,

Dennis

I had an idea to make robot skin out of "blister" type keypads. These are the thin, flexible keypads used in wet/dirty environments, or on cheap consumer electronic equipment. I found a place in Shanghai that could manufacture them for less than 10 cents per square inch, and could put up to 25 blisters in a square inch. They come with an adhesive peel-and-stick back.

Advantages: 1. Cheap 2. Robust. Impervious to dirt, moisture. 3. Can bend around a cylinder less than 1" in diameter 4. Easy to interface to a CPU. The number of I/O pins needed is 2*(sqrt of the number of pads), but you can reduce that by using a multiplexer and/or demultiplexer.

Disadvantages:

1. Binary (off or on), does not measure degree of pressure 2. Resolution is "only" 0.2 inch 3. Can curve in only on direction (cylinder) not two (sphere). 4. Limitations on how it can be cut into different shapes.

I didn't go forward with the idea because the minimum order quantity was too high (I think it was a hundred) and I only needed a handful. But if others are interested, I could revive the project. Let me know.

-bob

The two most commons ways seem to be scanning a matrix of switches (e.g. a keyboard), or measuring the voltage through resistive foam. With the former, you have no pressure sensitivity but can sense multiple contacts. With the later, you can only sense one contact, but you have pressure sensitivity. Of course, you could create a matrix of resistive foam pads, but all the wiring quickly becomes far too labor intensive for any usably large pieces of skin.

Could you elaborate on your design? What material did you use to construct the skin?

Regards, Chris

Hmm. How about putting a resistive foam pad under the matrix of switches?

Is there any resistive foam made specifically for pressure sensing? Or do you just use the stuff designed for storing ICs?

Is there a good way to sense a touch using piezo-electric sensors?

-bob

The problem with this is detecting multiple contacts. Simple switches can be scanned in sequence, and their on/off state won't effect the reading of other switches. However, if you replaced those switches with variable resistances, the readings from multiple contacts would average out and would be very difficult to make sense of. So in that case you'd only be able to detect the X-Y coordinates of a single point of contact. This is basically how resistive touch screens work, but that's probably not what you're looking for. You could of course get around this by creating a separate circuit for each sensor, but then the wiring becomes unmaintainable and quickly drives you insane.

The foam used to store ICs is usually pretty hard to work with. However, there are a few manufacturers who sell pressure activated rubber, like

Generally speaking, it can be difficult to measure static forces with piezoelectric sensors, but some are working on it. Here's an interesting paper on the design of a tactile sensor using piezoelectric resonance

However, I don't know of anyone who's made a commercial sensor using this design.

Chris

Pressure sensitive resistors(such as foam) under a matrix of contacts could be a usefull sensor, but resolving individual points could be non trivial. Energising one row to scan while grounding all other rows should help, but you'd still be reading row N against other rows.

Yes, I suppose you're right. If you put a diode at every junction in the matrix, you could scan the resistive components the same you would with normal on/off contacts. I believe this is the way a number of researchers are planning on implementing "robot skin" with the recent fabrication of flexible transistors.

Chris

Isn't this basically how a resistive touch screen works, though? One difference in the touch screen could be that the membrane isn't flexible, but that's not needed/desired in that application. It's actually pretty hard to design a transparent substrate for a screen that also has decent resolution. They've done pretty well with it, though resistive screens block about 25% of the light.

On the latest touch screens they deduce pressure by the distribution of the pattern (fingers flatten out; in robot skin the skin is compliant, so it's the same effect but in reverse) and force can be deduced by the timing relationships of the pattern. A sudden tap creates has a pattern where all the points are produced nearly simultaneously, whereas in a softer touch the pressure pattern is made over a period of a few more microseconds.

Of course if you're applying for a patent you can't say anything now, but I'd be be curious how this invention differs from resistive touch screen technology.

-- Gordon

Sorry, responded to the wrong message there. I meant that for Dennis!

-- Gordon

Dennis, this is totally off-topic, but how are you going about patenting this idea? Have you hired a patent attorney or are you doing the leg-work yourself?

As with the others here, I can't wait to hear more about what you're doing.