What would you pay for flawless occupancy detection?

[Quixote_1]

May I remind a few people here that this is a mini market research thread to help me to determine how much I can spend on building each of my devices, NOT a "let's figure out what the invention is" thread. Thanks. You're welcome to start yet another "effective occupancy sensing" thread if you so wish.

I don't see value in knowing how many people are in a room. And with pets and kids, 99% is simply impossible anyway.

You mean impossible like flying machines?
Motion sensors are the budget choice for this type of problem -- very cheap but annoying and inconvenient. I used to use motions for occupancy sensing but got sick and tired of always having to reactivate the motion sensor manually. It just made things even less convenient than not using any automation at all.
There are many uses to knowing how many people are in a room. One example I can think of off the top of my head would be when you are having a shower and your girlfriend walks in to brush her teeth or vice versa. How do the lights and music behave? How long do you need to wait for your system to realize that there is no one there and more importantly, if that time is low, how often does it shut the lights off while you're doing something that doesn't require enough movement to trigger the motion sensor?
Lastly, do you absolutely require a PC running all the time to figure out what to do when there is motion detected?

One of my goals in creating this solution is to eliminate most if not all of the headaches associated with stitching together some half-assed "occupancy sensing system", while making it affordable. While it may not be for everyone, there are a lot of other cool and convenient things that are not for everyone also. I know many people that scoff at the idea of having a smartphone, but I would never want to go back to a regular mobile phone again. That's just me.

As for identifying who's who in a room, Kinect can do this, at least on the 360 - I don't know if the PrimeSense APIs have any built in middleware for doing the same thing.

I can put to rest any notions that I may reverse engineer a Kinect and try to sell it as my own product. Somehow I think that in a legal battle against Micro$oft, I may not fair so well. But hey, if you want to (and can afford the time and money) to hack them for your purposes and then set up a 360 with a Kinect hooked up to it in each room, then by all means, go for it. More power to you.

 

[ver0776]

"You mean impossible like flying machines? "

I mean impossible as in A + B = 897, Tell me the value of A (without a variable in your answer (897-B ) is not acceptable as it is still an expression not a value.

I have a L-shapped room, it takes multiple sensors to "See" the whole room, how is 1 of your devices going to provide 99% accurate report of how many people there are when it can't see the whole room?. Yes it is a trick question. Just like a statement "1 Sensor will tell you who is in the room 99%"

I have to use multiple motions to cover the room $8 x 2. A $200 device would just become a $400 device to cover the same room? Will 2 of your sensors know how to work together? "Sensor 1 sees person A & B, Sensor 2 sees person B & C" How many people are in the room?...

Maybe my road blocks are rare, blah blah, but I promise you they will destroy 99%.

Write Brothers did fly, but not 99% of the time =)

Perhaps a more accurate statement like "99%+ count of occupants in a rectangular shape no larger than 100'x20', with adequate lighting and clear line-of-sight to all people, and they can not be wearing extreme clothing like coats and bizzare hats"

I am not trying to suggest alternatives, I am trying to say the price is based on ability to deliver, and prices of alternatives...

Vaughn

 

[Quixote_1]

"You mean impossible like flying machines? "

I mean impossible as in A + B = 897, Tell me the value of A (without a variable in your answer (897-B ) is not acceptable as it is still an expression not a value.

I fail to see your analogy. Forgive me if I seem rude (that is not my intention), but I think that it may be a little bit arrogant to assume that you understand everything that is possible. I think that most would agree that just because one does not understand the technology behind something, does not mean that it is outside the realm of possibility. Personally, I would be skeptical if someone told me ten years ago that we would have flat televisions that display 3 dimensional video for under 3 grand, but I wouldn't claim steadfastly that it is not possible and will never happen.

Now, I'm not claiming that I aim to revolutionize the home automation industry or anything, but I think that we all need to keep an open mind about where the industry is headed and what is possible, especially since it is relatively new. X10 may have been around forever, but it has only been recently that the general public has shown a keen interest in this sort of technology (probably because of the latest advancements).

I have to use multiple motions to cover the room $8 x 2. A $200 device would just become a $400 device to cover the same room? Will 2 of your sensors know how to work together? "Sensor 1 sees person A & B, Sensor 2 sees person B & C" How many people are in the room?...
Maybe my road blocks are rare, blah blah, but I promise you they will destroy 99%.
...
Perhaps a more accurate statement like "99%+ count of occupants in a rectangular shape no larger than 100'x20', with adequate lighting and clear line-of-sight to all people, and they can not be wearing extreme clothing like coats and bizzare hats"

I don't really care to give away trade secrets or elaborate on exactly how the device will work, nor do I have a working prototype at this moment. As I stated, this is nothing more than a market research thread. I will say that I aim to create something that requires no more than 1 device per room (unless you live in a gymnasium -- my apologies if that is the case). I can also promise that you can wear as bizarre a hat as you wish.
Lastly, I promise that I will not make false claims of its accuracy once I have my prototype and have tested it sufficiently. If I can create something that has an error 4 or 5 times a year and corrects itself without any noticeable hiccups, I think that I can righteously give it a 99% accuracy rating. Try doing that with $8 motion sensors.

Thanks for your perspective.

edit: spelling

 

[ver0776]

I am not asking for any details. And I gave my input $45 for the occupant count, $75 if they tell me the temperature too, $95 if they report analog light level, $125 if they can do all that and tell me how many dogs are in the room. $200 with IDs of known occupants and all other features. Identity recognition is the only thing that will revolutionize home automation logic. I would get a second job for that, well a 2nd paying job.

I love innovation, man, props to you if you accomplish something. I have seen so much heartbreak too.

Cheers,
Vaughn
OpenSourceAutomation.com

 

[Quixote_1]

I am not asking for any details. And I gave my input $45 for the occupant count, $75 if they tell me the temperature too, $95 if they report analog light level, $125 if they can do all that and tell me how many dogs are in the room. $200 with IDs of known occupants and all other features. Identity recognition is the only thing that will revolutionize home automation logic. I would get a second job for that, well a 2nd paying job.

I love innovation, man, props to you if you accomplish something. I have seen so much heartbreak too. Tons of money invested into this industry, few making much. Take the AB8SS, how much would you pay for it? I loved this product, supported it, and was very sad to hear the news of his manafacturing woes and the folding of the company.

You may think me negative, but I could be saving your life right now, keeping your retirement and family together =) It is generic advise because you know what you are doing better than me, I don't know anything. Just warning you not to base your life's fortune off of a couple hard core peeps that "say" they would give you $200. I dedicate my life to HA and I don't make anything from it. I am not a competator or anything, I would love to see your product on the market. Between MS and ASUS and all the clones that are about to jump into this CVP market, I would just be cautious. I predict CVP units under $100 in the next year, with code flying out after them.. Even if you don't use that tech, you have to price against it..

Cheers,
Vaughn
OpenSourceAutomation.com

I appreciate your input and your concern. I understand that, as with any business, there is a risk involved. I'm not going into this with stars in my eyes. Although I am cautiously optimistic, I am not trying to get rich from this (though it would be nice, of course). All I really want (beyond my business plan) is to make a comfortable living while doing something that I love and offering something that I genuinely believe in.

As much as I love CocoonTech, I won't be basing my entire business on what I've learned in one thread here. I'm just trying to get a general idea of how much I should spend on components, enclosures, etc. and how long it would take me to actually start making a profit. Seeing as most of the members here are quite knowledgeable and experienced, their input is highly valued -- significantly more than the average home-owner.

Interesting ideas for the additional features. I'll be focusing on occupancy for our first product, but who knows what the future holds.

Thanks.

I'm curious, why did you edit your post and remove half of it? I was in the process of replying when I noticed that.

 

[ver0776]

I'm curious, why did you edit your post and remove half of it? I was in the process of replying when I noticed that.

I thought the bottom part was a little off topic. I ramble on sometimes. I also have a habit of using Submit/Edit instead of Preview hehe. Didn't think you read it yet =)

Last tip on the product (any product), publish an API or some sample code so systems like mine can support it without spending more time on it then it took you to invent it =) Like when Insteon gave up on their SDK, I gave up on them. Hello Open-ZWave... I would pay twice as much for something with an SDK, not that end-users would, just my personal value added by SDKs.

It is why I am a big Phidgets supporter over Adrino & Hobby-Boards, etc. Phidgets sets the bar high when it comes to supporting their products with code and API's. I think I own everything they sell because of my appriciation of this...

Vaughn

 

[Quixote_1]

I thought the bottom part was a little off topic. I ramble on sometimes. I also have a habit of using Submit/Edit instead of Preview hehe. Didn't think you read it yet =)

Last tip on the product (any product), publish an API or some sample code so systems like mine can support it without spending more time on it then it took you to invent it =) Like when Insteon gave up on their SDK, I gave up on them. Hello Open-ZWave... I would pay twice as much for something with an SDK, not that end-users would, just my personal value added by SDKs.

It is why I am a big Phidgets supporter over Adrino & Hobby-Boards, etc. Phidgets sets the bar high when it comes to supporting their products with code and API's. I think I own everything they sell because of my appriciation of this...

Cool. I found the entire post to be pertinent. I started my reply but then my girlfriend showed up and I got distracted. Had to finish the post this afternoon.

Would sending an ASCII string to a preconfigured IP address or set of addresses including specified ports be sufficient? Seems like the most versatile solution to me.

The Phidgets were looking good to me as well, but I am hoping to avoid having to run a PC 24/7. Many others seem to be shifting in that direction as well. I can put a media PC to sleep most of the time, but when you are trying to monitor things like occupancy or, as you mentioned previously, temperature, light levels, etc., I think that the system should be continuously active.

 

[ver0776]

Would sending an ASCII string to a preconfigured IP address or set of addresses including specified ports be sufficient? Seems like the most versatile solution to me.

Yeah, just depends on what you send. The W800RF for example has a fairly complex rs232 protocol (for me =)... Tons of Bit flipping, hex manipulation, etc. If you have a complex protocol like that, source for the client that interprets the bit streams is really really helpful, API or not. (A Cocoontech member helped me with source for it years ago, or I probably would have quit writting HA software back then)

If your message is "Unit=1, Occupants=2" or "0110", I think we can parse it out on our own =) Again, with no idea how your product works, it is just stupid generic thoughts of mine... I know my partner found a cool XBee USB adapter the other night, I only had 1 question; "where is the source" =) Maybe I am just gun shy because of NIGHTMARES like Insteon. I can handle a little protocol work though.

Vaughn

 

[rismoney]

With a thermal imaging camera (maybe eye in the sky) you could get a reference baseline an empty room, and use image recognition techniques to determine the occupants.

Then using similar technology that picassa uses to sort images based on identity, you may be able to build an algorithm to positively identify facial patterns (distances between eyes for example) within thermal images.

GPS? not sure if its down to the meter... Lasers, Sonic/audio stuff?

Doable yes, Difficult - Hard, Under $200? nah I'll spend nearly half that filling my SUV's to get to the parts store
What would make it a lot easier is if you implanted RFID tags in all your guests ass-cheeks.


Your best bet is to file for the patent for your device, and share the technology. Then sue anyone who uses it.

 

[Quixote_1]

What would make it a lot easier is if you implanted RFID tags in all your guests ass-cheeks.

lol! Dammit! You figured it out! Well actually I was going to opt for the less painful approach of RFID suppositories. It would also be a great conversation starter when having dinner parties.

On a serious note, face recognition will definitely revolutionize home automation once it is accurate and does not require a dedicated PC just to crunch the numbers/ process images, but I really wonder how many people will be comfortable having cameras constantly trained on them, especially when they are in the bathroom or bedroom and the devices are all networked to the security system/internet. Sounds like a hacker's wet dream. Also, you might have a hard time convincing your girlfriend/wife that it is just an occupancy sensor and you are not in fact making dirty videos of her.

 

[ver0776]

lol! Dammit! You figured it out! Well actually I was going to opt for the less painful approach of RFID suppositories. It would also be a great conversation starter when having dinner parties.

On a serious note, face recognition will definitely revolutionize home automation once it is accurate and does not require a dedicated PC just to crunch the numbers/ process images, but I really wonder how many people will be comfortable having cameras constantly trained on them, especially when they are in the bathroom or bedroom and the devices are all networked to the security system/internet. Sounds like a hacker's wet dream. Also, you might have a hard time convincing your girlfriend/wife that it is just an occupancy sensor and you are not in fact making dirty videos of her.

Dude, you are so spot on. Kinects and stuff are all going to be fun and loved until people hear there is Nanny Cam software for them... I will never shoot up in a house that has an XBox or PS3 again! Wait, did I just say that?

But even worse is the blindness people have to Phones and Laptops. It is so hard to hide a camera, but now that people are oblivious to the cameras built into laptops and netbooks that you could be recording them right in front of their face and they would never think about it. Even before cameras, I would tease people in the Warrooms at work when I would walk back in, I say "I hope you weren't talking about me, my laptops mic was recording.... They all freak, I tell them I was joking, but anyone else with one of the 20 laptops in the room could do it for real and they should be careful...

And with phones, that person who always seems to be texting could easily be recoding your rant about the boss. I like to think I was always a good person, but I say some crazy sh*t sometimes and have a twisted sense of humor, someone recording me could make me look like satan... So as people wake up, I don't think they will want to see any inside cameras.

I am waiting for the market to get bigger before I start the wide-spread panic though.

So dude, if your device does not use a camera, or camera looking lens, etc, that would add some more value to me. I don't like scarring away the girls.

Vaughn

(Yes, I have now said Dude 3 times in a single post. I realize I may lose some forum privilages, but it is one of those nights)

 

[BioSehnsucht]

I can put to rest any notions that I may reverse engineer a Kinect and try to sell it as my own product. Somehow I think that in a legal battle against Micro$oft, I may not fair so well. But hey, if you want to (and can afford the time and money) to hack them for your purposes and then set up a 360 with a Kinect hooked up to it in each room, then by all means, go for it. More power to you.

Well, you don't have to do any hacking - just use freely available software APIs from PrimeSense or the other interface packages people have developed, and connect the Kinect to a computer... but I will grant you the physical shape size and shape of the sensors make them difficult to use directly. The upcoming devices that lack the color camera and only depth sense from ASUS and PrimeSense will be smaller, though.

 

[SpeedDemon]

Hi all,
I have a great idea for an invention that could solve all of our occupancy detection woes. I'm just wondering what the most you would pay (hypothetically) for a device that would 99% reliably detect not only whether a room is occupied, but the number of occupants as well. I'm not sure if I could build this for cheap enough so I'm just trying to get an idea of whether it's worth putting the time in to develop the product or not.
It would talk to the Elk so that you could keep track of the occupants using counters and trigger tasks based on occupancy and the number of occupants. What would you pay for a wired version and how much for a wireless version?
Thanks for your input.

How much would one pay for an advanced motion detector that would talk to an Elk?

My answer is $0.00, since (please, someone correct me if I'm wrong) the Elk is worthless in time-sensitive situations of triggering lights etc without imposing a huge 1 or more second delay from when the sensor is tripped.

Imagine you wired a motion sensor at the top of your staircase so that when you or your little kids come near the dark stairs late at night, the stairway lighting would turn on so you can see the stairs and prevent potential safety issues, and/or have your path down to the first floor lit up. Imagine similarly wiring a motion sensor for your dark garage/garage stairs too. Then imagine the Elk took so long to trigger your lighting after the motion sensor tripped, that you could walk almost halfway down the staircase in the dark before the lights actually came on.

HEY!! You just imagined my house!

But I'm not bitter. No sir!!

Substitute "HAI" for "Elk" in your question, and pretend I hadn't bought the Elk or remained blissfully unaware of its deficiency, and I'd say I'd have paid $150-$185/ea for maybe 2-3, or $45-50/ea for 18-20. If over $50/ea, my thought/buying process when I recently built my house would have been that I like the idea of the accuracy & multiple-occupant tracking, but to truly utilize the system to its fullest and track the number of occupants, I'd have to buy so many of the sensors (and pay corresponding wiring installation charges) for coverage that it would quickly move into the realm of "way too much money for a fun but totally unnecessary gadget."

I would have paid the $150-185 for 2-3 to serve as "critical sensors", as it would be worth the premium over the "regular" detectors I bought if these sensors had better accuracy plus wouldn't lose track of someone who stops moving for a while (say, while they are sitting on a couch staring at a tv). If the price per unit was low (~$45ea) then I might have sprung for a high quantity to get the coverage I would need to support occupant tracking. I would not have bought wireless, I wouldn't want the hassle of replacing all the batteries or risk any reliability issues with sensors located far away from the receiver. But with building a new home, it's an easy choice to hardwire everything.

It's sorta all or nothing if you really want to utilize the sensors to the fullest, right? You need full coverage for an entire house or you will lose track of people, and tracking occupants is one of the biggest reasons for buying the higher-priced sensor.

 

[Del91574]

We've already done similar to this on many installs using logic equations with our panels and installing Sure-Action Pulsors and processors. Works flawless, can be adjusted to account for/ignore pets and toddlers, and is invisible once installed. Only downside is it won't work on a slab.

 

[damage]

Well, you don't have to do any hacking - just use freely available software APIs from PrimeSense or the other interface packages people have developed, and connect the Kinect to a computer... but I will grant you the physical shape size and shape of the sensors make them difficult to use directly. The upcoming devices that lack the color camera and only depth sense from ASUS and PrimeSense will be smaller, though.

rumor is microsoft is going to release an sdk for kinect http://goo.gl/3Zn7v