[nzlug] Embeded problem - Dairy

[Andrew Errington]

I'm not an expert in farming, but I do have a lot of experience in embedded 
systems.

This is the sort of thing I would *love* to get involved in, except I am 
currently in Tokyo, not Christchurch.

I think your proposed solution is great, for two reasons.  One, it is simple, 
and two, it works in the way that the client imagined.  The alternative (with 
RFID tags for every cow) is not so bad an idea either, but it is expensive, 
and the client has already identified a problem for the milker/operator to 
locate and record the tag.

I would go one step further and suggest this idea (which requires no PC or 
Linux either).  It might not be such a good idea, but I believe it would 
work, and would be simpler.  The main area of simplification is the RFID 
system- basically there is no need to look up the ID in a list.  Either the 
cow has the tag attached (because the milker attached it) or it does not.

So, instead of a bracelet, imagine a 'sleeve' maybe 20cm long, with a large 
enough diameter to encircle a cow's leg (I know nothing about cows by the 
way) and a velcro or elastic seam so that the sleeve can be put round the 
cow's leg.  The sleeve is made from retroreflective material (the stuff that 
is on traffic cones and reflective tape for vehicles etc.).   Somewhere in 
the cow run there is an infrared emitter and detector.  The emitter sends out 
IR light at the height that the sleeve would be if it were attached to a 
cow's leg walking past.  The detector is next to the emitter and detects 
reflected IR light coming back.  When a cow walks past there is not enough IR 
light reflected from the cow's legs to the detector, but if the reflective 
sleeve has been attached, there will be.  The detector triggers a timer which 
opens the gate for a preset time period.  Presto!

Disadvantages are that the IR sensor assembly must be kept clean, and the 
sleeves must be kept clean, but milking sheds have to be kept clean, so 
perhaps dirt and, er, mud is not too much of a problem.  I estimate about $10 
for hardware (not including the gate mechanism).

There are disadvantages of the RFID system too, but you won't find them all 
until you try them.

HTH,

Andrew


On Thursday 27 December 2007 07:01, Simon Lyall wrote:
> I was at a Christmas do and talking to someone there who was asking the
> computer people for feedback. I'm not an expert in the embedded area (or
> farming) so I thought I'd ask here.
>
> The problem is that when dairy farmers are doing milking they will
> sometimes (perhaps once per day) spot a cow that needs looking at. It will
> appear sick or otherwise have a problem and needs further attention. They
> want to mark it so that can look at it later when the milking finishes.
>
> However this happens when they are in the middle of the milking and in the
> process of attaching the cups to the udders of the cows (from below) so it
> is hard to interrupt this and separate the cow or even climb up and find
> it's ear number.
>
> The idea they have is to somehow mark the cow and then as it is leaving
> the shed an automatic gate will trigger and divert it into a separate
> area.
>
> One of the farmers has talking to a vendor but his solution was to
> computerise and give special eartags to the whole herd and the milker
> would have to use a keyboard to type in the number of the cow for it to be
> picked out. Cost was around $70k which is a lot for simple requirements
> and having a keyboard etc a bit complicated.
>
> An alternative idea they are are thinking would be to attach a bracelet
> around the cow's leg which would send a signal to activate it gate. This
> would only take a few seconds and shouldn't cost too much. They were
> asking what hardware/computer setup would be required.
>
> The system I pictured would be something like:
>
> * Bracelet has RFID ( not sure which standard, somebody suggested
>   alternative magnetic system) in it.
> * RFID sensor a few metres before automatic gate
> * RFID sensor attached by USB to gumstick type small PC
> * Simple (couple of pages) program notices when one of matching RFIDs
>   detected
> * Program sends signal to standard automated gate via another USB port.
>
> Total cost say $1000 for the hardware, a day or so to develop and whatever
> the gate and it's interface cost.
>
> Other stuff:
>
> * The timing of the gate can be a bit aggressive since it doesn't matter if
>   the cow on each side of the intended one gets caught also. The farmer
>   can easily sort them out later by looking for the bracelet.
>
> * Some sort of simple control (USB device with a couple of buttons?) to
>   change the timing might be good.
>
> * Power in the shed is pretty good so it doesn't need to be super embedded
>   but it has to be fairly rugged.
>
> My questions for the group area:
>
> 1. Does the above seem like a good solution?
> 2. Does a couple of thousand dollars seem about right?
> 3. Does anyone know a company/person who could develop such a thing (the
>    farmers are in Canterbury so down there might be best) ?
>
> Feel free to email the group or me directly.