We have virtual switches, virtual rooms, virtual buttons... and they are critical and necessary...
The world needs Virtual floors.
Imagine if you could mount a device, 2 feet off the floor that created a virtual 'floor' that motion sensors would not see past! The dog could walk down the hall, and not trigger things! The robot vacuum could run and not cause changes to lights etc.
I feel a 'field' could be generated that interferes with the motion sensor in such a way as to stop the sensor from penetrating. How? beats me. I'm just an idea man.
Does/did that work? Maybe I'm overcooking it - I have 2 sensors that are most affected and if I could just 'calm' them down it would be powerful... but still... hmmm. Love to see/hear more about this topic and possible solutions!
BTW, that is the now obsolete Iris 3326-L1 made by CentraLite ...for my money, the absolute best motion sensor to hit the market. Great design, speedy, and durable. If Bruce ever wanted to market child devices for HE, that would be my nominee.
To some extent you can shape the beam using baffles of plastic or cardboard to mask of areas where you don't want detection. Damned hard to do it across an entire floor as you described.
I agree it would be useful (having two small dogs).
Your best bet might be using a camera for motion detection and then using the masking capabilities that most cameras have to define the "dead zone".
Another idea comes to mind - Camect does a pretty good job of detecting various kinds of objects. If you based your motion sensing on a camera system like that you might be able to exclude events that included objects such as cats, dogs, and roombas. I have not tried to identify a roomba with mine since I use it almost exclusively for outdoor cameras. It does a pretty good job of identifying cats, dogs, squirrels, foxes, etc.
Another vote here for Camect or other camera AI that can reliably detect “person” as an object. This attribute can be integrated with Hubitat via the Camect Connect community app. I have been using it reliably for about 1 1/2 years for a wide variety of security integrated use cases with my Hubitat.
Setup beam sensors like the ones that @mike.maxwell made and place them at a height that matches your dog and roomba. Then use rules that run motion lighting if the upper sensors are triggered with the conditional that the lower sensor should not be triggered.
Or do like @marktheknife suggested, and change the mounting angle of your motion sensors. I have done that in a few rooms.
The beam sensor modification works well. I use electrical conduit bushings (EMT bushing) as the shield. They are usually a perfect size and come in a bag of 5 or 10 for not much money. You can get them in 1/2" and 3/4".
This company spent a lot of years perfecting what you are asking for, first in the proprietary wireless alarm space, now available as a ZigBee module that will pair with HE and work with the Generic Zigbee Motion Sensor Driver. Sensor positioning according to instructions is important.
Just make sure you get the right model when Ebaying it.
Maybe detecting someone getting out of bed would be easier, rather than trying to get smart with the motion detection, unless you also want to have the same feature during the day. If you detect someone get out of bed and leave their room, then you enable motion detection outside the bedrooms....?
I've been watching the replies to the thread with great interest and thanks so far to all that have responded.
While my idea was a virtual floor (which I still think is a doable thing) it seems that others have resolved the similar problem with a dual sensor solution. Either a 2nd unit or a smart camera AI.
The camect is stupid expensive for me for a dog detector so it's not viable. I already have BI and unless I can learn a way to use it, I don't find that a real solution for my needs.
The 2ndary sensor at low height. If we explore this maybe you gents can elaborate a bit?
Let me describe the trouble spot and add a quick picture.
I have a hallway. There is a sensor above the door at each end pointed directly down the hall from either end.
There are 2 overhead lights also at each end. My current setup is if a sensor triggers the light at the other end goes on. In this way light in front of you is created.
the GF gets up at 6 and begins her day. I wake easily if light happens. She traverses the hallway but closes the door (not all the way) so as to stop light from going in the bedroom.
The dog gets up and follows her. He pushes the door open and bam. I'm up.
using a second sensor - how finicky would the rules be? You set a rule for 'if dog sensor goes off' to disable the 'if human goes walking' rule? Wouldn't the sensor above the door pick up both human and dog at a further distance?
(you cant see it well but red circle is where one PIR is - the camera making the image has a sensor right below it...)
Out of curiosity, what is the latency for this approach?
(Turning on lights with motion is sometimes more sensitive to additional latency than some other security use cases.)
Latency with Camect varies somewhat in my experience. Camect being local is quite fast, however, I have some of my exterior lighting connected through the Cloud (some original wall switches are GE Cync which utilizes wifi and are accessed in HE via Alexa). Because of this, my outdoor motion lighting based on Camect object recognition may take 1/2 second to up to just under 1 sec in my experience due primarily to Cloud service speed, not so much Camect speed. Fortunately, I have fiber internet connection so for exterior motion lighting this is not a problem in my use case.
The OP was looking at just making sure that his/her pets did not set off an interior motion alarm. In that application, I think that object recognition slight lag is not a problem (at least in my personal experience). However, if you want “instantaneous” response utilizing object recognition solely for motion lighting (especially for interior lighting), then for some, even a 1/2 sec lag may be too much. I have my Hubitat locally integrated with my DSC security panel via the Envisalink Integration so my interior motion lighting is handled locally via hard wired motion detectors wired to my DSC security panel. Response is of course, virtually instantaneous because of this, and is thus better suited for interior motion lighting (when used in the Disarmed or Stay Armed mode) or for interior motion intrusion detection (when used in the Away Armed mode).
The OP also stated that Camect was too expensive just to use for his/her pets. I would agree if other easier methods can be utilized. However, I use Camect for much more, such as object oriented security measures (where only persons would set off exterior perimeter security warnings etc), package and mail delivery recognition (Camect can distinguish between USPS, UPS, Amazon, DHL and Fedex vehicles), front and back door announcements (but only when it is a person, and not an animal wandering by), etc.
So long story short . . . (too late I know, lol) . . . response is pretty quick in my experience, but whether it is quick enough depends on your particular use case. In my case, Camect AI meets my needs to turn on exterior lights and to trigger a verbal response to warn away “persons”, but to not do this every time an animal walks by, although when I did have lights come on to “warn” away some curious animals (coyotes, mountain lions and deer) they just stopped momentarily like “deer in the headlights”, then just went about their business, lol.
Hope this helps.
