If a machine were to sharpen a knife, what "rules" would it follow?

I

Indlons_duded

Joined
Nov 27, 2010
Messages
3
Hello everyone,

I use a sharpening stone that is fixed at 20° off vertical and you bring the blade downwards and towards you (slowmo downward slash) along the stone.

I am trying to think of a way that a machine could successfully perform this sharpening activity, ideally better than a human. WTF I'LL SLAP U. I know, I know. There's not a machine out there that even comes close to craftsmanship. But I needed something interesting to do the project on.

And there could be.

The idea is not to make a sharpener that people would buy, its to think about what it would take to perform a better operation.

  • How many degrees of freedom (DOF) would be required?
  • What "rules" should be enforced by the control system?
  • Is it more important for the robot to be told where to go or how hard to push?
  • How to compensate for flexing / twisting in the blade?

The device would need to alternate sides of the blade for each stroke. Also, it needs to be able to follow the sharpening "rules" for basically any blade profile.

I am thinking at least 5 DOF. Below is the basic idea so far.

v1GQO.jpg


Not focusing on motor sizes or friction requirements, just assume we have the motors and bearings to make anything happen

Also, assume that as long as a sensor exists, it can somehow be implemented. On the right of the photo are two mighty accurate distance sensors.

I am happy to discuss what math i've got so far, I'm just now arriving at coordinate system definitions. If this discussion moves forward I'll share what i got!


Brian
 
Seconding hardheart's question. As to your question about dofs, you should only need four for a crockstick-type, assuming that the blade can be clamped such that the entire cutting edge is along a single vertical plane. The dofs would be two translational and one rotational in that vertical plane, and one translational perpendicular to that plane. Oh, plus one rotational around the major axis of the blade if we're dealing with double-edged or swedged blades.

The primary rule I'd implement would be keeping the edge at the point of contact perpendicular to the direction of motion/major axis of the abrasive. Given that lighter contact creates cleaner/sharper edges, I would think that force would be more important than displacement, and that the minimum forces involved would negate the need to worry about flexion for anything short of a long fillet knife.

Kind of curious that your first post is so in-depth, including industry-grade CAD. I'm guessing you're (studying to be) a manufacturing engineer?
 
The idea is young, the device is drawn as what i figured would be needed to replicate what i do when i sharpen a knife with my sharpener. I'm sure as time goes on its complication will come down fast (hope). This machine works to keep the blade edge flat against the stone with the blade flexed any (reasonable) amount.

Of the many ways to approach this, I started with what I do with my hands.

Also, one of the goals is to be able to accept any knife within a certain size range including flexy kitchen knives as well as stiff folders. Any blade profile. I see Tanto blades causing problems...

Which brings me to another point --> since the activity is so repetitive, The controller might be able to implement some sort of learning function. Say there's a spike in the force graph from the previous stroke, it could adjust the motion profile on the next stroke in a way that might reduce the force spike at that section of the blade.

The idea of light strokes many many times applied in just the right way makes me think that a mirror finish on a edge might be possible

@Bob I am studying to be an engineer in Colorado. Colorado is great :thumbup:
 
Mirror polished edges are pretty common, use an abrasive particle below 3 microns in size.
 
The first issue is the SM like set-up, if you draw the knife straight back and down you only create a straight line. Great for a straight blade but incorrect for a blade with curve.

When a blade has a curve like most do you must follow two angle axis. The first is obvious and that is edge angle, the second and most neglected is curve. Curve can be seen by laying the knife down on a piece of paper and with a pencil tracing the edge, the resulting line will be the curve.

Following the curve in sharpening is usually accomplished by raising the handle as the blade is stroked down the sharpening media. If actual measurements were taken I bet the angle of the X and Y axis through the belly to the tip would get pretty complex.

A more ideal set up would be to have the knife clamped by its spine with a free moving abrasive CNC controled. As for the abrasive I'd go with a 3 step grinding, sharpening, polishing process finishing around 1-3 microns.
 
how do companies like spyderco, bm or kershaw already do it? you could do it like that, just use more steps with finer abrasives.
 
Factory edges are ground freehand on belt type grinders.
 
You must log in or register to reply here.
Back
Top