ACCELLEROMETER FOR CLOTHES IRON

I think an accelerometer for a steam iron is an excellent idea, I and most technophiles would buy one instantly if marketed.

 Once you know where it is going you can then control its direction (with AI possiblye) and get it to do the ironing!!!

Some good links

http://www.dimensionengineering.com/accelerometers.htm
http://video.google.com/videoplay?docid=-312220290238860703

This is a prototype I am in the process of producing which adds certain types of vibrations such as ultrasonics to the iron soleplate which we have found have benefitted the process of ironing but I believe if these vibrations could be varied in power/intensity in relation to the direction of motion this benefit could be maxmimised and there would be much more benefit available.

Thank you PAKHTER.

Out of curiosity, would this benefit be great enough to be noticed by the consumer?
If so, how would the consumer notice it?

I would think that optical mouse technology would be more practical than accelerometers for determining direction of motion.
Acceleration would need to be integrated, which means that any errors would accumulate.  You'd also need to detect rotation, and distinguish gravity from motion.

We had the first prototype (without accellerometer) tested by an independent test lab in UK, and they ran a comparison to a good quality steam iron and found 23% faster ironing and 69% easier to move around so the benefts are enough to make a big difference to consumers if it could be further optimised.

Hi Ozo. I had a thought about the optical mouse technology, but even that seemed quite difficult to implement in an iron. Any ideas on how to port it into an iron?

or Doppler motion detection might be even easier.

Sounds ok but again how?

Project Natal technology ??

Natal technology (from what little I know of it) seems somewhat overkill for the iron even if the availability was not an issue.

All fun aside, the best implementation should have the least interference from external variables eg an optical position sensor can get dirty etc.
Current solid state accelerometers are easily capable of detecting the gross direction changes involved in ironing.

Also OZO's comments involve some inaccurate accumptions:
<<<<Acceleration would need to be integrated, which means that any errors would accumulate.  You'd also need to detect rotation, and distinguish gravity from motion>>>>
(1) In this case, acceleration does not need to be finely integrated enough to worry about accumulated errors. To be commercially viable, a max of two axis transducers would be allowed. If it detects motion in one direction it triggers the ultrasonics unit along that axis until the next gross direction change.
(2) Gravity is irrelevant. Ironing takes place mainly along the x-y plane.  Gravity exerts along the z-axis which has no component along perpendicular axes. Even if the ironing board was slanted slightly within human operational tolerances, the z components would be minor compared to the gross directional changes.
(3) Very little rotation is used in ironing techniques because it increases the risk of ruffling. Anyway, an x-y accelerometer system can detect this as both an x&y movement and trigger both x&y ultrasonic transducers accordingly.

NOTE:
more precisely "an x-y accelerometer system  in the nose of the iron".

Minor accelerations over time produce significant velocities.
But considering the possibility of being able to instruct the operator in using the device in a manner least troublesome to the sensor technology,
perhaps force sensors on the handle would be easier still.

Sorry ozo but I need to address your insistence on sticking to an inaccurate datum:
You are missing the point. We do not need fine integration;  we are not calculating positions, we are only registering a gross change in direction.
Also your phrase "over time" is a misleading assumption. There is usually at most 2 seconds between direction changes.
A 10hz polling frequency is sufficient. That approximates to 20 polls per ironing sweep.

Regarding the force sensors in the handle, yes it would be easier to implement but more expensive than the accelerometer because it would require a 'floating' handle with more component complexity and a greater probability of warranty failure. (on second thought, a flexible handle with the strain gauges glued to the inside, would suffice although the extra cost might negate the advantage)

A simpler implementation of the 'floating handle' would simply require rugged microswitches but the clicking would be very irritating.

Pakhter, just playing about with the microswitch approach, you could integrate the microswitches in the baseplate-housing interface. There is sufficient friction in that a change in direction would drag the baseplate sufficient to trigger the relevant microswitch.

I still think the accelerometer approach is the best. It is a black box approach with no extra moving parts, no casing redesign, no extra assembly complexity and less probability of failure.
The question is the ultrasonics. They will require expensive components unless you are only using eccentric cams.
Also how whould the users hand be isolated from the vibrations because there is a H&S issue gere.

Vibrations that could be felt by the user could be felt by an accelerometer.
10hz polling that happened to catch the ultrasound in the same phase could look like a continuous acceleration in one direction.
One possible way to avoid that may be to turn the vibrations off during the samples.
Although there may be ways to fool an unsophisticated accelerometer implementatiomn we may not have to worry about that if we can assume that the operator will not want to fool the accelerometer and avoid doing anything that might confuse it.

A good point. The apparent g changes from the ultrasonics could exeed gross direction changes.
The industry approach is to control the input bandwidth and pattern filter the trend. Mission critical accelerometers on gas pipeline emergency valves can distinguish between earthquakes vs passing lorries.

1/100 g, about a 1 degree tilt, for 2 seconds gives a velocity of 20 cm/sec.
that seems not insignificant.
And even if the direction changes, if the acceleration continues, the next 2 seconds produces an additional 20 cm/sec.
You might filter out everything under 1/4 Hz, but if the surface you are sweeping over is uneven or malleable, you may also have rocking motions near the frequency of the sweeping motions.

Yes, g*sin(1 deg) approximates to 17cm/sec2 but that is irrelevant because the  deceleration component purely from the user's change in direction could be as much as 320cm/sec2 (Calcs below). That is a factor of 16.

Please consider the following constraints:
(1) The g component in the tilted working plane does not influence the terminal velocity because the user brakes the motion to their working velocity so it rarely exceeds 80cm/sec. ie the g component only assists the user in accelerating to their working velocity. The  deceleration component purely from the user's change in direction could be as much as 320cm/sec2   (changing direction from 40cm/sec to -40cm/sec within 0.25secs).
(2) Velocity is not the metric used to calculate the direction of the vibrations.
"Rate of change in velocity in a given direction" is the metric.  This is not ballistic motion determined by gt^2. If a sweep was 'uphill' at the constant user velocity, and then reversed downhill to the constant user velocity the g values contribution to the velocity change calculation would cancel each other.

20Hz integration should suffice to trend a sweep and recognise a slanted working plane from the deceleration differences between directions although that runs the danger of positional calculations (whose position I oppose).

Uneven surfaces will introduce z axis motion. This is perpendicular to the x-y plane and any angular component contribution are theoretically not appreciable. The iron is not tilting up a hill and flopping down the other side. Any ironing board rough enough to introduce bumps should be tossed.  

Taking your point literally, we could always use a three axis accelerometer and brute force the problem.

JUST A THOUGHT: Does the g-component actually matter at all? The ultrasonic vibrations are bidirectional so if we are moving in the x-axis, the vibrations are transmitted both in the x and -x directions. So what if a slight g component assists the x direction and counters the -x direction; we will still be transmitting both ways.

So whats the end result can this be done and if so how do I make a start, would I have to go to a specialist firm to work this out?

> could be as much as 320cm/sec2
could be as much as, but sustained for 2 seconds, it would exceed most peoples arm length
so much of the time it could be less.
> so it knows which direction its moving  
says that velocity is the metric we are interested in deriving.

> Any ironing board rough enough to introduce bumps should be tossed.  
Even if you iron on gauge blocks, ironing over a seam could easily introduce 2 degree rocking

You could require that the user always use maximum acceleration to keep the error down to a factor of 16,
if tilting does not exceed 1 degree,
and you could probably mechanically filter ultrasonic vibrations, so 20Hz integration could suffice.
(although, at 80cm/sec, a 4cm periodic texture could cause aliasing of a biased signal)

an optical mouse sensor, or Doppler sensors still sound conceptually simpler to me, but I don't know what the relative costs would be.
Filtering to limit velocity as suggested could be done in software, and should not affect the implementation.

I wasn't the one who said
> There is usually at most 2 seconds between direction changes.

But an even easier implementation may be to fix the vibration direction on the iron,
and instruct the user to always orient the iron in the direction of the sweep;

<<There is usually at most 2 seconds between direction changes.>>
LOL Ozo, I thought that you were a precise physicist.
You misinterpreted that statement because you missed the word BETWEEN :)
The direction change itself is less than 0.25 seconds.
An ironing sweep at most takes 2 seconds (=at most 2 seconds between direction changes)


<<But an even easier implementation may be to fix the vibration direction on the iron,
and instruct the user to always orient the iron in the direction of the sweep;>>
A reasonable point and I did originally consider it. However some ironing techniques do require sideways movement.

BTW I'm only having fun here.

this for much of the 2 seconds, acceleration will be considerably less than 320cm/sec2

Compared to a true 0g acceleration with no direction changes, a measured 0.01g acceleration seems a significant error.
One way around this may be to ignore accelerations less than, say 0.05g.
Again, this makes it easy for the user to fool the the iron into vibrating in the wrong direction,
But maybe this can be considered a feature, not a bug, since the user can choose to trick the iron
only when a vibration in a different direction is desired.

There will be a whole new ironing language like Tom Cruise having an epilectic fit in Minority Report.

I dispute that OZO should get the points. I negated his suggestion.
I provided 2 suitable solutions, clearly explained the logic and negated any obhections.

I dispute that OZO should get the points. I negated his suggestion.
I provided 2 suitable solutions, clearly explained the logic and negated any obhections.