[OpenTRV-dev] Occupancy sensing, was intro

Damon Hart-Davis EMAIL ADDRESS HIDDEN
Mon Sep 30 13:41:45 BST 2013


Hi,

Please excuse some savage editing of subject line and below.

OpenTRV occupancy is still a lot more primitive than I'd like, but one technique I use a fair amount (sometimes it seems like only tool in my inadequate stats toolkit) is interquartile ranges of values observed over at least a day or preferably 7 days.

So, for example, when background luminance is the the lowest quartile having measured for at least a day, I deduce that people are not using the room or at least not active.  That's reasonably robust and doesn't depend on absolute levels.

A similar thing might work for even doppler: *relatively* little activity may be interpretable as equivalent to none or lack of occupancy.

But the stuff you're talking about sounds good too.  Good novel sensors and good signal processing / stats are a winning combo, I hope.

Rgds

Damon


On 30 Sep 2013, at 13:29, Adrian Godwin wrote:

> The problem with Doppler is that I want to detect occupancy without movement, eg sleeping. This isn't easy but we're experimenting with some 16 pixel thermal cameras. These might be able to learn environmental warmth and distinguish it from bodies.
> 
> On 30 Sep 2013 12:03, "Bo Herrmannsen" <EMAIL ADDRESS HIDDEN> wrote:
> ...
>  
> Hmm... about sensor... is that not only true for IR types?...
>  
> /bo
>  
> From: Adrian Godwin
> Sent: Monday, September 30, 2013 11:57 AM
> To: Closed list for developer discussions
> Subject: Re: [OpenTRV-dev] intro
>  
> ...

> On Mon, Sep 30, 2013 at 10:42 AM, Bo Herrmannsen <EMAIL ADDRESS HIDDEN> wrote:
> Hi
>  
> ...

>  
> From: Adrian Godwin
> Sent: Friday, September 27, 2013 12:27 PM
> To: EMAIL ADDRESS HIDDEN
> Subject: [OpenTRV-dev] intro
>  
> Hi,
> ...
> 
> I was at the Connecting Homes even in connection with an energy-managed house project. This also requires good occupancy sensors and we're looking at a crude thermal imaging camera. We think there might be a case for making a ceiling-mounted occupancy sensor with light and temperature sensors and I'm wondering if this could share some development with your plans.
> 
> -adrian
>  
> 



More information about the OpenTRV-dev mailing list