Engineering challenge

[Lliam20789]

I was wondering wether anyone had any clues or idea on how to build a robot/machine that can open a newspaper, close to centre? The trick is the newspaper is on a pile of 13 and needs to be able to open each one before moving it along.

Thanks in advance,
Lliam

 

[papaof2]

Back-of-an-envelope analysis:

1. Is this a "do one stack" proof of concept or is it for production?

2. Are the stacks of newspapers in a fixed position such that the robot must move along the stacks or are the papers moved to the robot?
- how will the robot move (guidance, power train)?
- how will the robot be powered?

3. Locate stack of newspapers
- known distance away from the robot?
- known distance from previous stack?
- all papers in stack are oriented the same way?
- all papers are within some tolerance (+/-) of each other?
- - i.e., smooth stack versus random pile

4. Locate edges of newspaper
- IR, sonar, light/dark sensor, microswitch?
- able to determine edge that opens?
- (already known if orientation always the same)
- save locations of sides for later use

5. Measure thickness of paper
- move paper slightly and measure distance from top of upper paper to top of next paper
- divide by two to get midpoint distance
- do this once and save for other papers in this run
- this should be part of the robot's "training" sequence

6. Move robot arm beyond the open edge of the paper
- distance depends on length of the paper lifting/clamping device

(7 assumes a device like tongs with rubber tips)
7. Insert lifting finger (thin metal or plastic) into paper at midpoint
- lift slightly
- insert lower arm of clamp
- close lifting device on upper half of pages
- closing force determined by thickness & weight of paper
- clamp device needs to be thin, strong, and have tips that can grip the paper

8. Lift and move in direction needed to open the paper
- amount of lift probably 4x the thickness of the paper
- amount of movement for opening is a little less than twice the width of the paper

9. Move opened paper to desired position
- (could be part of opening sequence - item 8)
- (need to include in "training" sequence, or have separate "stack finder" arm)
- (save for next paper in this run)
- release clamp
- raise arm
- move back into position to process next paper
- if last paper processed send request for more (or wait for "papers available" signal)
- determined by count if always the same number of papers
- based on thickness of paper and original stack height if count varies

Some of the finding/placing steps can be done once and saved ("training"), assuming the paper size and stacking locations remain constant. Note that the opening device must be able to lift the weight of the paper in order to move it to another location. This will require some serious hardware, as the paper is being moved long distances - the robot arm(s) needed must be able to support and move their own weight plus the weight of the paper.

Depending on the locations of the closed and open stacks, the robot could turn between the stacks (stack locations at 90 degrees) or the opening/moving device could be mounted on a bridge-style framework (stack locations adjacent to each other).

John

 

[hippy]

How would you do it with your eyes shut using a pair of tongs in each hand ?

As papaof2 illustrates, there are a number of things you will need to do. You have to overcome each obstacle individually then put it together as a whole. It's not an easy task I suspect and quite an ambitious project. Perhaps over-ambitious ?

 

[Lliam20789]

Lets just say it's worth it.
It is a very basic production line, not a proof of concept.

Thankyou for all your feedback!
Ok to answer those questions,
-paper can be orientated sam way, or can be 12 one way 13 another.
-paper will be in a holding/alining system, (front and sides?)

I'm not sure whether to have two basic arms, or one complex one...
I'm hoping to only use servo's (standard and mini) because the robot only has to move 1 paper at a time, never a bundle.

Good idea to grip the paper as it lifts, I didn't think of that.

 

[Rickharris]

How would you detect which way the paper is?

From experience with trying to orient small objects (Biscuits) it is very difficult.

Often easier to do these things mechanically via some unique feature.

 

[BeanieBots]

Get yourself down to a print shop and have a look at how their kit moves and handles paper.
Most of the time it's done with air/vacuum and rubber rollers.
I can't see hobby servos being up to the task due to the distances you would need to move but it might be possible.
Certainly an ambitious task. Good luck.

 

[Katzenjammer]

I certainly wouldn't like to discourage you, Liam, but that doesn't sound like a task for a PICAXE. Or even for many cooperating PICAXEn. The reason being, they haven't enough memory.

Consider how hard the task is for a human even with all our advantages of vision, dexterity, nearly unlimited memory, and a 'chip' that's specialized for pattern-matching.

I think Hippy has captured the size of the problem very well: doing it with a pair of tongs or chopsticks with your eyes shut. I think I'd add 'while you're wearing boxing gloves, and are drunk'. If you want to increase the amount of feedback your robot gets (i.e., give it vision plus pressure transducers in its 'fingers'), then you also need to handle that feedback in a fine-grained way. Which would increase your code by (finger-in-the-wind guess) a factor of 10.

 

[Matt_C]

I say never say "Never"!
It is probably beyound the scope for the forum to help as a whole project but how about breaking it down into smaller projects.
You will need a few PICAXE's.
Try to walk through all the processes that are needed and try to think how much you can do with each PICAXE and you may need a large PICAXE as the overall brain or you may even need a PC or some small computer as the brain.
It may be that the PICAXE's just do to donkey work!
I can also see the need for a lot of sensors to even detect basic things like paper orientation assuming when folded it's not square but then you could get round that and find the highest part of the stack assuming the following -

Does the paper fold dead flat at the fold or is it a little raised?
If it's raised you could use a light sensor and possibly a laser (for confined beam) to find the center. Start at the top of the stack and work down and test for the biggest gap and that must be the middle assuming you know the standard width of the paper when folded.
Then insert say 3 thin metal strips (2 at one end and 1 at the other end) pull the paper off the stack. 1 metal strip then presses down while the other 2 pull up to open the paper. You could use finger like devices to replace the metal strip to provide a good grip. The fingers could move away from the fold a little to make it easier to open. The opener fingers will need to move in like an arc when opening. Just watch how you open the paper when your reading it.

Do you have any speed requirements or is it just a case of 'as fast as possible'?

However, I think you could do it!

 

[moxhamj]

It would be a fun challenge but I wouldn't underestimate the cost. I know someone who builds this sort of equipment - eg Kraft might come to him and say they want a machine to wrap a block of cheese. Lots of pneumatics and servos and sensors - a cheese wrapping machine is about $200,000. If you got something even doing 1/10th of this sort of complexity I'm sure you would have a career mapped out. This chap I know always has a late model Jaguar.

 

[manuka]

Does the Jag. smell of cheese? Just teasing! This reminds me of the early 1950s boom days when a bale of wool would buy a Jag. Tales abounded of farmers saying " OK-here's 2nd bale to cover another Jag. for the wife". Here in NZ fruit sorting "robots" have been in use for many years, but their costs are often $$$$$. I've an old school mate growing apples who still swears by his ultra reliable DOS/486 driven one from the early '90s.

The real mechatronics crunch relates to BULLET PROOF RELIABILITY, with perhaps the Rev. Ed sweet sorter a good example- non uniform M&Ms may gag it. Proof of concept is one thing, but ensuring a smart machine is able to respond to real world variations may blow project costs out by orders of magnitude. The Wright brothers showed D.I.Y flight was possible, but they only initially managed 12 seconds (?) aloft. How much has the infamous Airbus come to - $ ½Billion ?

 

[hippy]

"How Stuff Works" and "How it's Made" TV programmes enthral me. I never cease to be amazed by wire or rope twisting-come-winding machines; now who got their head round that design to achieve it !

In my youth ( 1970's ) I lived in Manchester and we had school trips to working cotton and wool mills and the Grauniad [sic] newspaper when still using hot metal presses. Amazing. Worst experience was a working visit to a paper mill in Kent in the middle of summer; the noise I could live with, the heat and humidity I could have done without. Have you even noticed there's always a pub next to a paper mill

 

[Matt_C]

I think the pub is to do with liquid cooling

 

[BeanieBots]

I agree, however, the background music in "How it's Made" can send you insane if you watch two episodes consecutively!

As for wire winding machines. The most impressive piece of kit I ever saw was a transformer winder. The toroidal type transformers with a one piece core. Not so much the ingenius method of winding but the shear speed at which it was done without breaking the wire.