Carnegie Mellon's Synthetic Sensors Project

This looks like an interesting and potentially game changing approach to sensors. If a decent UI and engine can be built to make such a sensor system really usable to could definately change the way we go about sensing the smart home environment.


A cool device, but a total geek bubble as far as understanding what is really needed from home automation. Take this:

And that, Laput says, is the end goal of a smart home: building an environment that knows more about itself than you do.

Because that is so not the end goal of a smart home.

The end goal of a smart home is in the actuators, not the sensors. It’s improving the convenience, safety, accessibility, power utilization, and cost management of everyday activities.

It’s not about knowing whether the lights are on: it’s about turning the lights on when they are needed, and off again when they are not.

So, yes: an interesting idea to have one device provide tons of sensor data to a home automation hub. But it’s just data. It doesn’t give the hub the ability to make any of the devices in the room do anything. And it’s the doing that ultimately matters. :wink:


But think about how we could use the sensor’s reading as triggers for our already installed smart devices!

Right, the sensor information could definitely be helpful to the automation hub. It’s just the sensor by itself doesn’t do what the article says it does – – it doesn’t connect devices at all, and while it can measure activity on dumb devices, such as water flow at the kitchen sink, it doesn’t provide any way to turn that dumb faucet on and off.

So definitely a cool device, and this approach may well be adopted for future home automation systems, it just is only the beginning, not the end as far as delivering home automation. :sunglasses:

Hate to do this, but… choice is good :slight_smile: And my choice as to whether lights are ‘needed’ or not is gonna be different from yours - and in fact, from my wife’s choices too.
So someone striving to create a truly smart home cannot ever know where to set that threshold. Therefore, the folks designing sensors cannot concern themselves with end-user criteria for the right amount/duration of light, the right water temperature, the right audio volume level, etc.

From the standpoint of a component designer, his statement makes a world of sense. He ultimately wants to provide a home that knows, before you do, that your hot water heater is on the verge of failure. So that you don’t end up with a flood-damaged basement. He wants to provide you with a refrigerator that knows the coolant is beginning to leak out the moment that starts, so that you don’t lose $400 worth of food. He wants to provide you water valves that know they are off, so that if there is any water flow when all valves are off the system can pinpoint for you exactly where the water is flowing. Etc.

Sure, I think the sensor itself is a cool device, and that kind of sensor may well be key to much future automation. But it’s not the “end goal of a smart home.” That would be thermostats, not thermometers. Obviously every thermostat needs to have thermometer information, so you want both. :sunglasses: But the information alone is not enough.

As far as choice, absolutely: there have to be multiple different ways of defining The parameters that go into “needed,” as well as override methods.