[00:00:06]
>> Up to the higher level where our family myself and Lance and. Saying well what are the kinds of objects you might produce how often they are the energy to perform some computational task and bring all these things together so with that kind of overview of what we're trying to do in Cosmos I want to introduce the last example who is a recent hire at the University of Michigan he did his Ph D. work at University of Washington working with Josh Smith who many of us know better couple of times you.

[00:00:40]
Talked about his work I think around this summer prior to becoming a natural Ph D. student was working with bins out of hours service but you know growing that up and yeah I'll answer the back of the electrical engineering but from my perspective was very interesting is he's not just interested in the whole circuit design he's interested in what kind of applications that might support after finishing University of Washington he went to work for you continue to work for Intel research and then eventually went to Disney research in Pittsburgh and ultimately worked his way up to becoming the head of the lab there at his research and compost everything that they could possibly imagine so they decided to close down the lab it was no more.

[00:01:26]
And so the reef areas where nicely nicely side very nicely so I used to see if you could complete the mission there. And then fortunately for the University of Michigan unfortunately for Georgia Tech they will quickly be able to sue them out to go and join the family there has been for a little less than a year now and as he'll show up we've done a lot of work in harvesting and doing novel sensor developments to provide a wide variety of applications for any of you that are interested we're going to spend it this afternoon with Keira and the last number of people who know the.

[00:02:05]
Roger just from my. Elaboration that you and your partner. Well I think thank you for the great introduction so I have spent the last 10 years in industry more or less some very academic minded labs but generally my goal has been to let computers understand how we live work and play right and that's the general problem has ben that computers don't know that much about us so we think about our phone it only really knows where I'm supposed to be because I told it right and it knows generally you know where I am because of G.P.S. but otherwise it hasn't have any real awareness of what I'm doing or what I want I think we coming from Disney right I want I think Jarvis right I want that.

[00:02:54]
Iron man that knows what I'm doing in the moment and can help me do something and so a lot of my work has been a around that source of technology I think about cellphones is a great portal to connect to other people or things but again it's bad to understand what I want I think smart homes are kind of facing the same problem I have you know.

[00:03:15]
Refrigerators they're connected to the Internet for someone reason I don't know why that's good. But still doesn't the spaces really know what I'm doing to know what my family's doing isn't really helping us and so we kind of think about what's what's keeping us from that. You know we think about Mark Wiser's vision of you because computing we have sensing in all over the place and what are the things that are keeping that from happening so from my electronics background I can think to myself and say well we have compute it computes everywhere you get a microcontroller that has enough horsepower to launch people to the moon for under a dollar we have contacted city right so here's a wife I model for a couple dollars they can connect us to the Internet so we can talk to anybody in the world and sensors are cheap we have those devices they're everywhere but what's keeping us from having this ubiquitous sensing devices that really help area help us in the days.

[00:04:05]
Life in the answer really is for me from my point of view is power right we don't have ways to keep these things perpetually going and clearly the AI folks would say Well nobody's making smart decisions about what to do with that information well I agree but we will I will get them the information that's kind of what I've been working on a lot of ways so for all my work you'll see me trying to solve the how do I make computers aware of what I'm doing it was in the context awareness and how do I solve the power problem.

[00:04:34]
So I'll talk about 3 kind of research efforts I've had over the years one isn't battery free sensors one is more on the interaction side of how do you make devices that are where you get contextual where in is what we're doing and then kind of my core and more dual lexical engineering this just still love wireless power and I've been doing lots of things about how you send power one place to another through people to medical implants and all sorts of things OK.

[00:05:01]
So one of the one of the technologies that are found especially is begin my Ph D. was critically interesting is our Friday so for those who aren't very familiar how these large ID tags are like stick or form factor things you have are fighting readers and in the long range case you can power these things up to like 5 to 8 maybe 10 meters and if you're good right and so they're like these little micro computers or you can stick on things are always trying to do work right we're trying to compute the protocol the coming towards them and boot up and reflect some energy back to to send some data and what I really like about these is they don't actually send any signals they reflect signals us to somebody shining a flashlight at you that would be a reader and you can't shine a flashlight back you have enough power to make light so instead you just take a signal mirror and bounce information back and it's a very cheap way of sending data and how is it.

[00:05:52]
So that's the some of the advantages of our of ID And so if you think a bit further if you look at the computational instructions for jewels of seam OS or any. Logical mind you know microcontrollers microprocessors the competition power doubles every 2 years or so and we made the observation that if you look at the freeze transmission equation which is how you can send power from one place to another place as a function of time.

[00:06:22]
Then as a function of distance you find that OK So computation instructions bridgeable doubles every 2 years that means the distance at which I can wirelessly power something doubles every 4 years so it's like wow my research and our idea where I was going to get better just by standing still is you know that's kind of amazing So that's a that's a pretty good interesting trend for the same workload I'll be able to have double the range in 4 years and then another interesting kind of view of the world is your computational cost of sending data so here we have power costs and you need so you have power consumption versus bits per 2nd and you can see up here we have legs being wife by and Bluetooth and they're all these kind of conventional architectures up there and you have.

[00:07:08]
Got a university architectures which are still super heterodyne Seema sort of things and you could argue that they're all about in the same class because doing about the same thing one that saw it different is alter wideband So this is a completely different buy So here's any of these laws Barbara impulses and then you get them to alter it you get the backscatter here and here you see like 123 orders of magnitude better performance and bits per joule So that's like wow that's amazing and how you can do all sorts of things like instead of saving dady it's cheaper it is to actually transmit it so it has all these really big system in cations for electronic systems.

[00:07:44]
OK So while the Intel working with working with Josh we made this device so this was a our fighting tag that didn't have any batteries but didn't have any special silicon in it so it was a because we thought of the software to find our body tags so had less range so for denying meters depending on which version is still used backscatter but it had this complete my controller so now you could do all sorts of interesting things with it.

[00:08:10]
And so you know from my R. a point of view I had all this nice impedance matching in network an antenna stuff and I had this rectifier to harvest the 150 megahertz energy had power management and demodulation blocks and a full microcontroller on it so you could sit down an undergrad could reprogram this thing to be something different in about half an hours it's fairly straightforward temperature sensors and accelerometers and L.E.D.S. and water crystals so all the things you want from and print it primitive in that IT systems but no power no battery whatsoever so that makes really crazy systems that means you have a energy harvesting system that kind of looks like this you have some source.

[00:08:52]
Some find a source impedance so you have some power that's coming in an energy harvester and you end up like harvesting air you charge off your entire storage might be like on a 10 micro faired acid or and then you turn on your burning energy really quick analysis and you brownout right now how are you going to get any work done in this sort of scenario you have to get your entire process done in transmit it off in that time and then you get all these sort of like we're programming things really OK maybe you can check point one other time is like your your stack of cropped and now you have your pointer error and you're off somewhere else and so all these really interesting programming need bugging things started occurring so we've made a bunch of progress and just how you debug these things.

[00:09:32]
But in the result this is what you kind of get so here here's a wisp in there isn't our Friday reader it's probably a metre and a half away in this example and as I move this whisper around. You can see Saturn turning in kind right so no batteries it boots up Harbor says Our hearts are if energy boots up measure the acceleration of gravity downwards on the fly computes in our Friday packet with a series he sends that back and that's interpreted by our Friday reader and then application right.

[00:10:07]
So. So one of the probably most important things I learned from working at Intel and working with Josh is that I made this device and as a young engineer just getting my masters degree I was like Alright great now we're going to optimize the intent of. And he's like no why don't we spend like 9 months open sourcing this things I was like that's crazy I won't write any papers but we got an army of undergrads to help us and we ended up open sourcing this and giving away to I think 50 different research groups by now over 500 which has been sold or been donated so now there's a small community of people there working with these things building applications so that was really.

[00:10:46]
Informative for me as I would be came in a professional so I hope to do similar projects in the future so. Yes there's a whole wiki there's a working group kind of working on these sort of things we had a conference on it so some of the kind of work I've done in this area is we did.

[00:11:04]
Be an energy harvesting so we went and looked at trying to harvest and you know we're already waves that all around us and if you go look at that Moore's Law graph again you start seeing that the amount of power you need to do computation has dropped to the point where she is signal levels before we thought you needed power and now it's just signals and so we're able to harvest enough energy to run this little.

[00:11:23]
Temperature sensor from my 4 kilometers away and then we improved it ended in wireless sensor node from 10 kilometers away so I have kind of a big in town I guess of course but basically there's the N B energy around this is enough to do real computation work.

[00:11:39]
And. This was T.V. So we're harvesting from T.V. towers. We've also looked at cell towers wife by and in this as a power goes down your range goes down we did big loads so we started like a little microcontrollers fairly easy but if you have something like a camera that has to be on for a long time and has a lot of heart large amount of power Anyways we figured out ways to harvest energy and store on larger capacitors and now we have these full cameras that are able to take pictures and do image processing on board a very primitive but still so valuable.

[00:12:18]
And then like most people like me want most people but many people many of my peers lots of backscatter work or ambient back got to work so we looked at all the radio signals that are around us and you saw those as a medium of which we can back scatter information on to see Imagine you're in a lake you have water all around you and here you can use that medium to send data from one device to another device so we made a harvester or little hearts you know that would go from a megahertz up to 2.4 gigahertz use that higher bandwidth to send data back all right we're doing time.

[00:12:53]
Let's see skewed a little bit. So I had these wisps and they were nice devices research quality but still cost you know 50 bucks for one of them and you need a graduate student the 2 of them and they're fairly hard to work with and we wanted to think of ways of how do we increase the usability of these devices.

[00:13:15]
So one of the things we want to do is figure out how do we go and stick them on all sorts of devices to figure out what you're doing so. One of the observations we made is that if you understand the objects interact with they both reflect and define your daily activities so if I know what you touch that day I can tell what you're doing and provide you with better services you can think of things like elder care or life blogging for helping you complete tasks.

[00:13:39]
So in this sense instead of putting the sensor directly on the tag we started thinking let's use the communication channel as a sensing mechanism so our Friday tags talk to her readers talk to our Friday tags right and they report back there are some ID and now modern readers give you back like not only our society but also phase in the channel that you're transmitting and Doppler shift and you can use that as a sensing mechanism so now I don't have that before my controller and accelerometers all I need is a normal Our Friday tag and I can tell things like which tags are moving right or which tag is in the environment I get a snapshot of it's our local our environment so when it reports I know about the those 4 parameters for this tag which is different from that tag and I can tell things like the tag is moving I've touched the tag or put my hand near the tag and interfering with it and that lets us do some interesting things so this is one of my students graduated from university Washington and I'm reading all the tags here and then in another instance I can tell which is moving right just by looking at the changes in phases function time and i gave me some higher order information now I can tell you what he's doing right I can tell you that somebody drinking tea in this room.

[00:14:56]
We kind of dive into the signals a little bit this is a can sample trace of what you have seen so there's our society kind of still and then there's some movement and then it's still again but the phase isn't make a lot of sense sure it's a little bit more organized when it's still but then the less organized winds moving but you can unwrap as a function you take a 2 2nd window you can under up is a function of the channel and now makes a lot more sense these fall in these kind of pirating This is from 0 to 2 pirating and then you have these little plate face splitting where there's multiple bits on top of here and then they get pulled out when there's moving things are moving around you can take that and make some basic machine learning classifiers So we went through and said OK here's a motion in rotation so I put your finger across here and covering it and then you can do things like this.

[00:15:47]
So at Disney of course we cared about toys. So we hear for $7.00 to $0.10 you can go instrument a toy and give it some basic functionality you can put it anywhere on the device it doesn't really matter where it is it's just easier to put it there and so here is the real time classifier there's a 2 2nd delay and it's telling you the difference between different types of interactions and you can tell that out of a population of toys what you're interacting with.

[00:16:19]
And so I particularly like this because although it doesn't matter at all with the complex who the circuits are all you need to do is have something to back scatter and you can get all this this type of sense in your. Vehicle. Right so you are changing the R. environment of the entire space so you your reader is probably in the ceiling this example so your arm is over it so you're still getting some side signals but you just have to have enough threshold.

[00:16:48]
To classify we're looking for. Kind of along those those lines you can see that the classifications this is the confusion matrix and so when we say that something when something really is still we predicted to be still 92 percent of the time you might say these these numbers aren't that great for real time interaction but certainly for monitoring people's behavior this is the signals are particularly good.

[00:17:12]
So some of the interactions that I'm particularly interested now in was interested at Disney is can I start understanding what you're doing in your daily life so if it's 6 am and you get the cereal you get a bowl here we can see these classifications of these move interactions as a function of time and get the milk is pretty easy to understand what you're doing right you're making yourself breakfast but that's really important for an elder care situation where now I know that my elderly grandmother is taking care of herself she's eating in the morning I mean I remember my my one of my grandmothers if she had such bad arthritis she wouldn't eat unless she went out and we didn't know that and if we had known that we could have kept her living there in her home longer.

[00:17:52]
Years. There are not sure now that will this come down to a lot of our signals are differential So we're looking at the change in it not the absolute value and certainly what container is with water and it's going to have lower read range then attack this on a dry container typically we could make tax because if we have that we're all.

[00:18:16]
Measuring volume that you can do that then those people have done that as well so that's really possible to add other more static sensing. Wow. I don't think you'll be able to do that I mean we should probably look at it more in detail. I look at the other ways when I get moving vans can I then figure out what you're doing in that home so we just publish a paper we're trying to look at the difference between different cooking habits and can you tell the difference between such one chicken and lemon chicken and if you tap tag the right things you can tell the difference between those 2 meals being prepped.

[00:18:59]
That would help yeah you could but we go back to the West scenario we can we had we made a milk wisp that measure the temperature or measure the temperature and the volume level on the on the device so you can do more as that with our IF idea does become a challenge how do you have state when you're away from the reader so it sometimes have designer interactions actually take advantage of when your have power.

[00:19:24]
Yeah that would be great let's do it. No no no pressure no pressure. And kind of a little bit along those lines we did some printing right we're not as sophisticated as security is but we just got a. Posse here to go to PA Yes So we just use the silver and chip into a printer right so we can make the antennas and we have these the loop tags and you can put that loop tag on here the inductively couples and now we can print out all our intent a structure is an absolute sticker and make all these interactive paper sorts of things.

[00:20:09]
The sticker has a little piece of our little R 5 the I.C. in it still and it has a little loop and it's supposed to be for jewelry in the reed ranges like a few millimeters but then we can add it and actually couple into this intense structure and now we can have more and more control over the interactions with our body tax.

[00:20:26]
See have an example. So you can do things like Scott Hudson's classic button slider right so here you can make all these sorts of devices out of our ID. Tags and so some of the are all using different techniques some of them are like my contact some of them is a slide in your finger actually becomes the ground plane in completes the antenna some of them are you're just looking at the interference pattern as you move your hand around and likewise you can hook them up to other systems right so here you put our ID tag down and you can control the lights so you disconnected to feel guilt to.

[00:21:09]
All sorts of weird things. Yes OK. And then this is a test taking thing. So in the interest of time. So one of the challenges that we face with our idea is that this is what it normally takes to set up on our Friday reader you need computers in a reader and network and you what you really need is a post-doc that's.

[00:21:37]
Going to get to work then you know that's that's what's necessary and we made the observation that you know we have conventional light bulbs and now there's a whole bunch of volume in incremental lightbulbs right they've been replaced with L.E.D.S. in fact you can put wireless communication in them and sit B. and all that sort of stuff so we wanted to make our own light bulb so that we could easily just deploy our 40 readers we could put them everywhere in fact my mom could deploy them an important aspect of H.C.I..

[00:22:03]
So here you know making another joke about how many P.C.'s to turn to take eternal life. Yeah here the problem is you also have to S.S.H. into this one so it does become more complex but you can have a full our Friday reader now this is a little bit larger than a standard floodlight mostly because we're not mechanical engineers.

[00:22:26]
So just an anatomy of an arbitrary boat so here we had like a full one while our Friday reader and fans and Elliot E.'s and we had a trunk down or a with it. Inside of it which is actually pretty challenging and we now we have full appreciation for how much work goes into making L.C.D. light bulbs we don't overheat.

[00:22:49]
They spend a lot of time doing thermodynamics while we just put more and more fans into our light bulb. But as a result now we have a way to monitor infrastructure pretty easily. So in the previous examples were always instrumenting. Dynamic things like spoons and forks and bowls and cups and here we were trying to look at how do you actually an instrument the infrastructure so here we can get information about drawers and.

[00:23:28]
Doors and there's a as actually attempted tag that measures the temperature in the room and then when the thermostat tagged the reports back to the the set point that this is so now we can bring that more easily into the home. All right. Quickly go through this one in the interest of time.

[00:23:48]
In the vein of trying to help. Computers understand what you're what you're doing we started looking at E.M.I. spectrum and if anybody probably not in this crowd would if anybody tried to get anything FS F.C.C. tested this is a terrifying graph because what's happening is you spend about a year making your special device and you take it there and then they go to scan to see if it's F.C.C. compliant and here you have to be underneath the F.C.C. says I don't care what the modulation type is I don't care anything about that you have to be underneath that line when these 2 lines and now you're over it and I have to get spent 3 more months before you can ship a product you have to go fix this problem but that's the important thing here is that they don't care what that modulation type is underneath it so we went and hacked together this $10.00 radio receiver which is supposed to be used for a T.V. tuner and we turned it into a software defined radio naturae move the our friend and made a very fast dock from 0 to 3 megahertz and made into a wearable not allowed us to capture the raw.

[00:24:50]
R.F. spectrum that comes off of devices so here when you touch a phone you get to know my signature child across your body into your into your hand it can go from either hand doesn't really matter so now we get to get a snapshot of what you're touching and that seemed interesting in and it took us a little 2nd to realize how profound that was is because the E.M.I. spectrum that comes off devices is very unique per device and that's because the circuits underneath them are apparently different.

[00:25:18]
So for instance a toaster oven is far different than a fit or a do all drills or a math book they all have very different circuits and therefore the in my spectrum is going to be different. One. We did that study we looked so it depends on the tolerances of that manufacture so we looked at we tried to Philips light bulbs in compact fluorescents and those tolerances are very high and we could barely tell the difference we looked at Dell monitors had one monitors are exactly the same model in our office and we could easily tell the difference between them so some of that down to the item level is based on what's already in your database.

[00:26:05]
OK So the simple mechanism here is basically So you've got smart watch you're looking you have a dremel drill here you get Aurora dual baseband shift and then subtract out your background noise and if it's something's missing delayed if it's significantly different you can pass it off to a machine learning algorithm and here it was fairly simple he just uses for effect a machine classifier and was awesome here is like you know here we're like 100 percent right with the accuracy is really high again because we have a lot of frequency diversity right in this unlikely that these things are going to overlap because they're actually different devices and in this case we looked at measurements.

[00:26:49]
You know month won or whatever and then 2 months later we actually did the user study and so these profiles were consistent with function time and that makes sense because they're all frequency based like the oscillator your crystal oscillator didn't start drifting that's a function of time is sees the same rate and I'll show this one quickly just this is couple of quickly so here you'll see the classification results as the student goes and touches different things with C. So here's the spark of the gas stove that's the touch surface touch table underneath.

[00:27:38]
So good. So we drill Now 1st off you know 3 drill presses on. So you can you can tell the R.P.M. of your drill that's pretty easy you can tell you that way scale we can tell how much you weighed because Oscar is like some relaxation oscillator for the main mechanism so yeah you can also if you want to dive into the details you can figure out state information anybody know why the ladder was giving us a signal when i guess it's in a town exactly right so this it took a long time to figure that out right yeah.

[00:28:12]
Yeah exactly. We traded for 3. So this was it. F.M. country western station in Pittsburgh and we were under sampling that signal and that was dropped into our frequency range so it took us a while to figure out what the actual source was. And. I personally cannot. And the OK.

[00:28:48]
Yeah in this ladder is like bolted into this part of the lab and so it probably unbolted one you know one half of it would resonate differently pick up if we can see and stop working so there are static structures you can categorize. The way. There I mean that's I think music directory Hightower from Intel great he did place lab which was basically looking at.

[00:29:41]
Signal strength and why 5 N.F. am I think is well maybe now maybe not but he was looking at Signal strength and as a fingerprint for your location so yeah. Yeah. Yeah Lester you know it's a good antenna. So the signals propagate their cell in the near field right where they go through things pretty easily the magnetic fields and electric fields will profit across your skin they'll go through PASTIKA they're actually horribly hard to shield OK.

[00:30:27]
I'll skip through this because I realize over time but there's one more interesting thing that I am getting more excited about so I left work a lot on wireless power and so you have these devices that you want to run perpetually and we think about and Dr charging right this is like one day charging like I want to charge my Sonicare toothbrushes in one spot and if you move it all doesn't really work and then you have to be charging pad and from usability point right you can put it on the pad and yes you get to do it to these but I still have to I'm stuck here and what I want they want my phone in my pocket I want to charged as I walk around the room that's half Michael and this ended up being a bulk of my Ph D. thesis I started out doing this sort of stuff so here are these high Q. resident coils that are oscillating back and forth and they'll send power to each other so this is a former resident coupling where the can interfere weakly with each other transferring power here and transferring 60 watts about 90 percent are off to our efficiency over 3 feet or so so a younger better looking me.

[00:31:32]
I went to work on medical implants so I worked with about the heart surgeons from Yale and we made a fully implantable version so this heart pump is like an elevator this actually sold in the into the heart pumps for you and there's a wire that comes out and that was that's approved therapy right and then the problem is the wire causes infection leads directly to the heart so we replace that actually got the animal trials doing that and then tell us the same technique to make our sleep our laptops and cell phones we actually got a product group excited enough to pick up our work and so they made laptops that would wires the charge cell phones and all my 3 D. charging I was very excited about it you know 3 D. charging they got it down to about this much when you actually do practical engineering you got about that far and OK so I learned a lesson right and that is that this mechanism is only good for about a coil diameter at best right so you know how big oil's will go farther if you have a little coil that only goes whatever the coil amateur is and doesn't work well with asymmetric oils so when I got to Disney Well I was looking for something else and this is how you work in wireless power people say well then Tesla do that how's that compared to his work and it got me thinking so he had this idea that you could use the I own this here in the in the ground as a waveguide and sent a lot for manic energy all around the planet and it turns out that's an interesting idea and this is back in the late 1900 when no everybody thought it was insane to run cap or cables to every house around the country that was an insane idea so this was actually possible the problem was that he overestimate the connectivity of the honest here in ground so there just just wouldn't support a wave for that got me thinking.

[00:33:18]
And I said well I'm going to Disney and this guy asked me to start solving this charging problem where they had our cast members you know young 1918 year old were plugging. In their point of sale devices at the end of the day can you make me a chamber that recharges So OK I'll happily work on that so I made this chamber here and I decide OK I'm going to put a standing wave in this and see just can I just transfer power right so that's equivalent to putting a copper conductive in a sheet on the ceiling seen in the floor we put standing wave up on that and transfer energy.

[00:33:53]
So that worked well there was this and we published it and we could transfer energy anywhere in there and here's this is basically based on how the resonators. Here is in this particular mode there is high connectivity for magnetic field here in here and we show that we can transfer power everywhere in there but there was a problem and that problem was that due to the arrangement in a field electric field we could not you cannot put humans in there if you want over a WATT OK So the challenge is how do I change my relationship magnetic field or electric field and such a way that allows a human spin there so long story short is we had this idea that we wanted to control a current flow has the current flow around that metal chamber if we could have that current pass through a capacitor or other positive element that would allow us that change the resonant frequency just like we did with those of wire that I did for the for the stuff at Intel and I could lower the frequency to point where would still resonate but it would be safe and so we figured out that we could instrument this mode this resonator with this pole and put a capacitor inside and the resulting magnetic field would swirl around it.

[00:35:06]
OK so I wanted you know I was at Disney and I said OK well we have some extra space and I want to make them all small little model I want to make something big so OK so we need a 16 foot by 16 foot room and we're up to ceiling so 7 and a half feet and and so it's in this particular interest in sanitation there's metal all around except for this door you don't actually need as much metal we found out later but it lets you do some interesting things.

[00:35:34]
Oops come back and in the key trick here is there is a metal pole inside and you could cover it with dry wall or whatever you want to but there's capacitors on that pole that allow us to lower the frequency so I'll cut to the chase in the show you their result here so here is an amplifier we're going to send.

[00:35:50]
An R.F. signal at $1.00 megahertz into this room this is a coil of wire we're going to star oscillator in a pump this room full of magnetic fields OK they're going to back and forth and what's going to happen come back. Try one more time. Was going to happen is that the magnetic field is going to cause current to flow through the walls floors and ceiling up into that pole the pole the current going to go into those capacitors until it gets up a G. right and then that's going to drive back out of the capacitors through the wall causing the magnetic field that swirls around that one field will collapse causing current to try to fly through the wall so now you're storing magnetic fields in the room and electric fields that are harmful for humans in the capacitor oscillating back and forth so this is my postdoc Matt he has a loop a wire with an antenna is this showing where the magnetic field is in this room and so just about anywhere you are you can get some of this magnetic field.

[00:36:50]
He's brave. Is actually very. Good. And it works anywhere in the room except for the corners so we have a circular mode and we have a square room so a square room round pegs sort of scenario. For that because that would look like a fairly cage but we found we don't need as much metal as we thought we did and we have a big hole in the in the wall.

[00:37:16]
So here you walk anywhere he walks in the room it works you'll see that he does the trick here he turns the thing to make it stay on that's because you need to have your loop of wire be able to catch the magnetic field so you have some orientation problems right and we've since figured out ways to overcome that but it's an important thing to point out so you have to be orientated correctly but you can do more than L E D's So here we have you know fans we took out the batteries we put wires coils inside the case and rectifier and now you can turn the room on and actually have real fans going.

[00:37:50]
So you can do real work with it and so back to what I wanted the whole time I want my phone in my pocket to get charged but here's the problem when you do where it was power really well it's super boring right super boring. Right so watch this.

[00:38:05]
That's the you know the icon changed right. Right and so if you go and show that to a Disney executive they may not understand they want to say and why that's so exciting Thank you yeah yeah I learned that one time this just little hint. But yes so you got the thing now you can mouse yours we mean yeah.

[00:38:29]
So yeah we looked at different ways of making rooms that have wireless power everywhere let's see. You can certainly make 3 orthogonal coils so that you can have wireless power in all directions right we could also change the mode structure of the room so you can live or power anywhere.

[00:38:48]
And here's this example we can make it look like a living room you can have big pieces of metal in there those are metal bookshelves is long and so at these frequencies you don't get any current loss so you're not heating things up but. If you have a big big piece and all the magnetic fields are only just moved around it unless you cover up so much magnetic field it actually between the tank each in the room.

[00:39:10]
So fairly fairly invariant objects around the environment we of course look to look at efficiency and model this and so my postdoc is Hero out as well as equations and went from a closed form solution to figure out. The solution I like to measure things so that was my that was my mechanism and in there they both worked out.

[00:39:35]
I think the important thing here is to look at safety right so it's great to be cavalier about wireless power systems and how great we're going to change the world but if you don't you can't use them in the real world it doesn't really matter so the F.C.C. has to metrics that they typically look at start with so one is how mention how much electric fuel you're exposed to and volts per meter and the 2nd one is how much is your body being heated up so specific so for like one gram of body mass how much temperature does actually rise how much energy does it get and so it's amazing what you.

[00:40:05]
In find online there we found full high resolution M.R.I. scans of a human we asked our digital artists to decompose it take out the organs in annotate them will correct electrical properties of these frequencies we threw them into a complex simulation and then did the basics our analysis to figure out when when how bad is this like how about that So here's a basic plot here just is just the input power that we can put into the room so you know and then as a function of so that the area where we can operate is in the green region and what's particularly exciting here is if you were a 90 percent efficient you could transfer 1 point one kilo watts of power still safer humans in there so I would never do that I'm not to ask my postdoc to go in there bagus even if we're not officially at you know at 0 percent efficiency just room and so on $100.00 watts is still a lot of usable power even even $2530.00 watts is still a lot of usable power so that gives me some insight that there's actually some some room in this technology right if you're willing to instrument your room you can get power everywhere.

[00:41:11]
And like I said we did a bunch of work on changing the modes around pushing energy around so that you don't have that orientation problem and then we made sensor nodes that used harps and power but also use that communication medium or that what magnetic fields a commission communication medium so that the room could talk to nodes nodes could talk to the room nodes could talk to nodes so just another way to do it in your sensor networks right so without its want to go and of course I have lots of great collaborators all of these folks have been from.

[00:41:45]
Disney as well as Intel and some clubbers that you Michigan and universe washing so without question. All. This is a great question so you can touch the pole you're just fine I have not touched the top of the pole and the bottom of the pole. But in the reality you would go and put that into you know you have a pole right there right there is that there's a bulkhead over there so you can put drive all around it and we'd never notice or put it between walls and I have to I was barrooms the question.

[00:42:39]
Over. OK. I haven't I though I suppose that you would need a static field right so I think you're as closer to that with the sound I feel than I am. Right there. So that's that's a great question so part of this is what's the ecosystem compatibility so there's like she and then there's a 4 piece will a higher frequency so that she is at $300.00 kilohertz A for W.P.S. at $6.00 or $6.00 megahertz whatever it is.

[00:43:29]
One is that low frequency. And is very good for going low distances and that's what's in your i Phone in your in your Android phones your other one is good for being up longer distances. I would be worried trying to go too far low we started a study on this and we haven't finished it but if you go to load then you get a current losses everywhere and that's bad if you go too high then you start a reading people so there's a magic number in there on that chart where it is not sure where it is that hits the standards back very.

[00:44:02]
Very. Very. Very. Yes. I can show you some other slides but basically all the panels on the walls we we ended up taking off as they were we're deconstructing the room because we finished our mission as a Gregor pointed out. We could take up all the other walls in the currents going up through the walls so we had pillars I would imagine that the ceiling panels are probably more important right in the floor panels so.

[00:44:44]
Yeah yeah and we had a big we had a big you know panel that was missing so you have doors and windows I think the question is how do you how do you make a wall treatment that's easy to install without having to put metal everywhere so this is is it a conductive paying is it a wall paper do you have to buy a party or dry wall windows and.

[00:45:09]
I am talking to us she right now about some of those things I'm more interested in how does architecture change when you don't have to worry about running power everywhere it sounds interesting what does that architecture. When you don't have to worry about wires you know. Where this more configurable work spaces in your house and.