Not sure if this falls under robotics or home automation?
I have an idea to use an Arduino to control the direction in which a table fan blows air. The thought was first to use PIR to locate me and aim the fan in my direction as I moved but decided that would be too imprecise and would be confused if another person entered the area. ( It's my air...get your own )
What I think I will try to do is use IR sensors and a simple IR tag clipped to my clothing that emits a pulsed IR light in all directions. ( 555 timer, LED and coin cell ) The sensors would use auto tracking similar to solar collector following the sun to keep the fan pointed at me as I move around my shop.
I expect my first Arduino board and a few items such as a PIR sensor, a relay output board and a couple of servo motors with controllers to begin experimenting with.
As you will note if you read my most I immediately eliminated the PIR method for that reason.
Before I actually try this I will do some learning to get a feel for the platform.
Basically I envision a platform about 12" dia and about 1.5 to 2" tall on which any suitable table fan would sit.
My initial thought is to use perhaps 5 IR sensors/lenses arranged with one in the center position where the fan will "aim" with two on each side spaced to permit them to see a fairly wide area.
If I use a PIR for anything it might be as an occupancy sensor to simply turn the fan on when I enter the room and off again after I leave.
I found a few discussions on solar panel tracking. I need only one axis of course. When I get to the point of experimenting I may need some help learning how to program to tell Arduino to try to look at 5 inputs and aim to put the IR transmitter in the center detector. I have a pretty good idea what I need it to do but not yet how to get to that point.
I will follow up on this here later after I get my first Arduino on Monday and have a bit of time to learn about the hardware and software.
if the fan is oscillating you might put a ping ))) distance sensor on top of it and measure the distance in the various directions.
Make a reference table for the distances in all directions without a person.
Then it can do one sweep every X minutes. If no one detected wait X minutes
Otherwise set the fan in the right direction and let it blow until the ping ))) sensor sees the person moved.
Start over again.
Actually, PIR should work if you use 2 PIR sensors, and arrange them so there is a small
overlap of their receptive fields - out at the edge. Then, you will have a "binocular"
system with 2 wide peripheral pickup areas [approx 120 deg each, depending on the
sensors], and a common area of 10-degrees or so where the edges overlap.
When you walk into the room, one of the 2 sensors will pick you up, and the servo (I
presume there's a servo) will pan over until both PIRs are picking you up. Then, as you
move it will continue to follow you (you will write good s.w. for this, I presume).
Once it's locked on, it will stay locked on. If someone else enters the room, it won't
budge, as long as it's already locked on (assuming the s.w. is written this way).
As I recall, I did this 5 years ago using PIC software, but it's simple enough
to gin up. Just read the 2 PIRs and do some if...thens. The important thing
is adjusting the PIR receptive fields. Mine was easy to do, since the units
I used had essentially horizontal pickup patterns, rather than 2D patterns.
Sounds good. As a former alarm company owner I am familiar with masking PIR's to control their field of view. I would mask these to a horizontal fan only a few degrees "thick" across the room to avoid lights or anything else that might cause noise.
I will probably be a bit before I start this...right now I am working on extracting an engine from a welder for use on a hovercraft.
( Oh ... and I still have a web company to attend to when I have spare time !