Google developing “Brillo” Internet of Things OS based on Android

Google dives into smart homes with an OS for all your low-power gadgets.

Read the whole story

 

[vogelabv]

adurbe[/url]":3f5u4xyg]They already own and operate Nest. Will this mean Nest will switch to Android or will Google maintain multiple IoT platforms? Or... will my Nest be left to the wilderness?

I'd imagine IoT devices would be those that plug into Nest. Nest itself, operates as a hub and may always run a different base OS.

 

[Trandyr]

Ancaritha[/url]":wjvn47pv]I work at a fuel cell company. If we can design fuel cells that fit inside of hand-launched UAVs that run on 64KB of ram, I feel it should be possible to have a 'smart' light bulb with less than 64 MB of ram...

Of course it's possible. The issue is that when you are dealing with 64KB of memory, the software has to be specifically tailored to the device and it's independent needs. That means increased overhead in designing software and a lack of interconnectedness (sp?) with other devices (which is necessary for IoT to function properly). The whole point of the project would be to give developers who write IoT software a base to work on that already provides a lot of the necessary functionality for networking and communications out of the box. It decreases development costs and time while also preventing fragmentation.

At least, that's the hope.

 

[Trandyr]

Ali2000[/url]":2c4agr71]"Such devices need to boot up, use an SoC, handle input and output, and communicate over a network—all things the Linux-based Android OS is great at;"

That's such claptrap. You don't need an SoC for most of these IoT things; a microcontroller will work just as well and Android is a bloated OS if you're talking about " light bulbs, door locks, sensors, and whatever other crazy connected objects the IoT crowd dreams up".

Lets put this in perspective. The GE Smart Light Build uses 0.4W on standby (measured by someone on a forum). Not a lot you may think but that is about 50% of what it uses over it's life time. One of the large components of residential electrical demand are vampire loads, devices that are on all the time sucking power. SoCs and Android are the wrong solution.

Then what's the right solution? Honest question here. I'm curious what your take on it is.

 

[zaqzlea]

siddharthvader[/url]":36h9po32]

aimed at ultra low-power devices with as little as 64 or 32MB of RAM.

This made me smile, as the devices I work on have ~256KB of RAM.

Such devices need to boot up, use an SoC, handle input and output, and communicate over a network

You can this kind of thing in ~100KB RAM. A full blown linux based solution seems like overkill - a basic scheduler should be good enough. Especially as we are talking about things like light bulbs and door locks.

Agreed, even using the linux kernel could be unwise if the chip doesn't have an MMU (I know μClinux exists, but that's quite hackish).
The preferred solution IMHO would be a simple 'hard' RTOS (like FreeRTOS or TI-RTOS), which is just a glorified scheduler. Actually, most of the complexity will be found in the Ethernet (or WiFi) and TPC/IP stacks, and that can be mitigated with specialized hardware (external or inside the SoC).

 

[DeschutesCore]

iolinux333[/url]":1qehgitb]

DeschutesCore[/url]":1qehgitb]

Oletros[/url]":1qehgitb]

DeschutesCore[/url]":1qehgitb]

thegrommit[/url]":1qehgitb]Brillo? Do they not know what a Brillo pad is?

Brillo is also the Latin word for 'bright, according to Brillo. Not sure how true that is.

Brillo is not a Latin word, is an Spanish one

From http://www.brillo.com/history.asp

Loeb accepted the offer and in 1913 secured a patent for the product under the name Brillo® (derived from the Latin word meaning "bright.").

Long ago, they taught Latin in public schools. I guess that, like many brilliant bits of our society, they did away with superfluous stuff like that by the time you arrived?


http://www.etymonline.com/index.php?term=brilliant

I went to Californinan public schools in the late 80's. We didn't have any exposure to Latin aside from a kick ass science teacher in Jr. High.

 

[DeschutesCore]

Double post

 

[Roberto76]

Oletros[/url]":2fa3948r]

DeschutesCore[/url]":2fa3948r]

thegrommit[/url]":2fa3948r]Brillo? Do they not know what a Brillo pad is?

Brillo is also the Latin word for 'bright, according to Brillo. Not sure how true that is.

Brillo is not a Latin word, is an Spanish one

According to the "Dictionary of Spanish Language of the Royal Spanish Academy" the Spanish verb "brillar" originates from the Italian verb "brillare". So definitely not Latin, but could be also Italian instead of Spanish.

 

[weblionx]

I do wonder if a low-voltage low-power communications bus that could power up devices when needed would be better. Ignoring (unfortunately) the cost of the extra cabling, it would simplify many things and increase power efficiency by having a single hub that plugs into the devices that can wake them up and control them. All the complex processing gets done in the hub and each device has a serial connection (maybe multi-drop like RS485) that has some standard protocol that lets the hub discover the capabilities of each device. Even most of the lowest powered microcontrollers can do serial communications.

I could see it using low-speed networking over the power lines, but that seems like it would still result in the stand-by power draw you get with Ethernet or (even worse) with WiFi. I wonder if power line networking could be done in such a way that a different carrier frequency could be passively band-passed to turn on (and latch) the device's power supply. This way there's no stand-by power draw but the hub could still wake it up.

Any sort of higher-level networking seems like a waste of resources to get flexibility I can't see being used for most things you would have on the "IoT", and for those few items that do need it (anyone have an example?) they could communicate with the hub over WiFi/Ethernet instead, while leaving things like lightbulbs on a much more efficient setup.

</thoughts>

 

[NoMorePosting]

This will be an interesting race. Microsoft isn't behind for once.
(Just downloaded windows 10 IoT preview thingy from them for Raspberry Pi 2).

Of course it would be nice if they played nice together - Z-wave, Zigby, WiFi, Bluetooth are competing standards at the comms level, but at a higher interfacing level - proprietary all the way most likely.

 

[Chemical Tribe]

AdamM[/url]":27se0nx6]

thomsirveaux[/url]":27se0nx6]"Google wants to move in and clean up the fragmented mess by offering Brillo for free to OEMs"

I feel like I've heard this one before.

That's the same one that didn't have a very happy ending isn't it?

Sometimes when Google tries to swing in and clean things up I feel like they've never seen this XKCD.

http://xkcd.com/927/

So how's Google's form of Linux going vs. the endless bickering of what will replace X11 or which desktop or sys5 vs....

 

[NoMorePosting]

Roberto76[/url]":1vatwsaj]

Oletros[/url]":1vatwsaj]

DeschutesCore[/url]":1vatwsaj]

thegrommit[/url]":1vatwsaj]Brillo? Do they not know what a Brillo pad is?

Brillo is also the Latin word for 'bright, according to Brillo. Not sure how true that is.

Brillo is not a Latin word, is an Spanish one

According to the "Dictionary of Spanish Language of the Royal Spanish Academy" the Spanish verb "brillar" originates from the Italian verb "brillare". So definitely not Latin, but could be also Italian instead of Spanish.

Given all the Romance languages developed out of Latin, are you sure? tbh I'd almost aggree, 'Brillo' is almost too short for Latin ...

 

[Chemical Tribe]

DeschutesCore[/url]":6qip8dke]

iolinux333[/url]":6qip8dke]

DeschutesCore[/url]":6qip8dke]

Oletros[/url]":6qip8dke]

DeschutesCore[/url]":6qip8dke]

thegrommit[/url]":6qip8dke]Brillo? Do they not know what a Brillo pad is?

Brillo is also the Latin word for 'bright, according to Brillo. Not sure how true that is.

Brillo is not a Latin word, is an Spanish one

From http://www.brillo.com/history.asp

Loeb accepted the offer and in 1913 secured a patent for the product under the name Brillo® (derived from the Latin word meaning "bright.").

Long ago, they taught Latin in public schools. I guess that, like many brilliant bits of our society, they did away with superfluous stuff like that by the time you arrived?


http://www.etymonline.com/index.php?term=brilliant

I went to Californinan public schools in the late 80's. We didn't have any exposure to Latin aside from a kick ass science teacher in Jr. High.

Not knocking you or your comment, but your username, wasn't that a code name of an old Pentium 2 or 3? Interesting.

 

[fuzzyfuzzyfungus]

I'm having some difficulty imagining why you would want something based on android for 'IoT' applications.

The only variant I can imagine having any logical traction would be the (Oracle-enraging) notion of doing with Dalvik what Sun originally did with the JVM(had it running, more or less on bare metal, little or no OS, on a variety of embedded platforms, with the theory that people would be willing to buy somewhat more expensive microcontrollers/microprocessors if they could port their code more easily). I don't know what terms Oracle is offering the various mini-JVMs under these days; but if Google undercuts them, some sort of bare-metal-dalvik could probably move some units.

Aside from that, though, Android is more or less a pig: an entire embedded linux build, with an I-can't-believe-it's-not-a-JVM on top of it, along with a graphics system that has taken ages(and a lot of powerful hardware) not to feel rather sludgy and a sound system that still has giant delays in it. For an application that probably doesn't have its own UI, and doesn't need Play store compatibility, what could possibly make that more attractive than the just-basic-linux builds that run on basically all the routers ever(or something smaller still, at the expense of less familiar development)?

 

[gmerrick]

ugh... connected home. what a security nightmare.

The only thing I really want to see "connected" are my 240V programmable electric baseboard thermostats, so I can control all 20 of them from one program instead of having to go to each and every one of them and do it manually.

 

[kindakrazy]

siddharthvader[/url]":39gkc932]

aimed at ultra low-power devices with as little as 64 or 32MB of RAM.

This made me smile, as the devices I work on have ~256KB of RAM.

Such devices need to boot up, use an SoC, handle input and output, and communicate over a network

You can this kind of thing in ~100KB RAM. A full blown linux based solution seems like overkill - a basic scheduler should be good enough. Especially as we are talking about things like light bulbs and door locks.

it has to have enough code to make it difficult for developers to change it so it never talks to google's servers.

 

[Edzo]

Even if you're determined to use Linux for this, why go with Android over something like OpenWRT That's already suited to networking and less powerful SoCs?

 

[Edzo]

Even if you're determined to use Linux for this, why go with Android over something like OpenWRT That's already suited to networking and less powerful SoCs?

 

[Chemical Tribe]

fuzzyfuzzyfungus[/url]":2ym3ft4y]I'm having some difficulty imagining why you would want something based on android for 'IoT' applications.

The only variant I can imagine having any logical traction would be the (Oracle-enraging) notion of doing with Dalvik what Sun originally did with the JVM(had it running, more or less on bare metal, little or no OS, on a variety of embedded platforms, with the theory that people would be willing to buy somewhat more expensive microcontrollers/microprocessors if they could port their code more easily). I don't know what terms Oracle is offering the various mini-JVMs under these days; but if Google undercuts them, some sort of bare-metal-dalvik could probably move some units.

Aside from that, though, Android is more or less a pig: an entire embedded linux build, with an I-can't-believe-it's-not-a-JVM on top of it, along with a graphics system that has taken ages(and a lot of powerful hardware) not to feel rather sludgy and a sound system that still has giant delays in it. For an application that probably doesn't have its own UI, and doesn't need Play store compatibility, what could possibly make that more attractive than the just-basic-linux builds that run on basically all the routers ever(or something smaller still, at the expense of less familiar development)?

I doubt this will be using nothing of Android's sound/gui systems.

But as LP has shown, the issue was never hardware, just the way it did garbage collection, compiliation, etc.

 

[bthylafh]

siddharthvader[/url]":2kfpgjtu]

aimed at ultra low-power devices with as little as 64 or 32MB of RAM.

This made me smile, as the devices I work on have ~256KB of RAM.

Such devices need to boot up, use an SoC, handle input and output, and communicate over a network

You can this kind of thing in ~100KB RAM. A full blown linux based solution seems like overkill - a basic scheduler should be good enough. Especially as we are talking about things like light bulbs and door locks.

I suspect Google is trying to lower the bar to app developers. It'd be easier for an Android dev to write a Brillo app (assuming its native language is Java-ish or Go, or something else high-level) than an assembly program like I think the devices you're talking about would run.

 

[iolinux333]

Roberto76[/url]":12q1w0k0]

Oletros[/url]":12q1w0k0]

DeschutesCore[/url]":12q1w0k0]

thegrommit[/url]":12q1w0k0]Brillo? Do they not know what a Brillo pad is?

Brillo is also the Latin word for 'bright, according to Brillo. Not sure how true that is.

Brillo is not a Latin word, is an Spanish one

According to the "Dictionary of Spanish Language of the Royal Spanish Academy" the Spanish verb "brillar" originates from the Italian verb "brillare". So definitely not Latin, but could be also Italian instead of Spanish.

Verb? Back in my day public schools still taught that old fashioned noun/verb grammar thing. I guess they did away with that superfluous stuff before you got there? I've heard everyone gets an iPad now though, and doesn't have to experience the horrific sound of screechy chalk on slate.

 

[panckage]

[edit]

 

[WangChung]

All these people commenting, few of them actually read the damn article.

Google isn't talking about sticking the hardware OR software power of a Galaxy S6 into a toaster. It's a redesign of Android with a much smaller footprint that has just the basics of what the smart object in question needs.

How do you TL;DR five simple paragraphs? I even checked the Flesch-Kincaid score - 9.9. "High school sophomore" is really that far removed from people's comprehension?

 

[Ali2000]

Trandyr[/url]":2qc8w4da]

Ali2000[/url]":2qc8w4da]"Such devices need to boot up, use an SoC, handle input and output, and communicate over a network—all things the Linux-based Android OS is great at;"

That's such claptrap. You don't need an SoC for most of these IoT things; a microcontroller will work just as well and Android is a bloated OS if you're talking about " light bulbs, door locks, sensors, and whatever other crazy connected objects the IoT crowd dreams up".

Lets put this in perspective. The GE Smart Light Build uses 0.4W on standby (measured by someone on a forum). Not a lot you may think but that is about 50% of what it uses over it's life time. One of the large components of residential electrical demand are vampire loads, devices that are on all the time sucking power. SoCs and Android are the wrong solution.

Then what's the right solution? Honest question here. I'm curious what your take on it is.

I already mentioned that for the examples provided in the article a microcontroller is a perfectly good solution. IoT is an over-hyped TLA covering anything from automobiles to light bulbs and too many people talk as if there is one solution that will cover all of these, its as if they are looking for a single magic bullet to solve IoT. An SoC is appropriate for automobiles, a microcontroller which can come in and out of a power down mode in micro seconds rather than milliseconds is more suited to a Smart Light Bulb. ZigBee is used in LightBulbs, Wi-Fi might be more appropriate for other devices. In time the real truth of what Brillo is about will emerge but as others in the comments have stated, 32/64MB in a light bulb is poor engineering, for a refrigerator maybe. But the author didn't mention refrigerators, just light bulbs, locks and sensors and fell into the magic bullet trap. We should start with the use cases and not the technology.

 

[Ali2000]

WangChung[/url]":2w07szs5]All these people commenting, few of them actually read the damn article.

Google isn't talking about sticking the hardware OR software power of a Galaxy S6 into a toaster. It's a redesign of Android with a much smaller footprint that has just the basics of what the smart object in question needs.

I did read the whole article and I'm still saying it's way too big for what the author's stated uses are: "light bulbs, door locks, sensors". You don't need Linux for that, you may need it for other IoT stuff but not for a light bulb. That is what people are reacting to.

 

[Oletros]

Roberto76[/url]":10oijpxu]

Oletros[/url]":10oijpxu]

DeschutesCore[/url]":10oijpxu]

thegrommit[/url]":10oijpxu]Brillo? Do they not know what a Brillo pad is?

Brillo is also the Latin word for 'bright, according to Brillo. Not sure how true that is.

Brillo is not a Latin word, is an Spanish one

According to the "Dictionary of Spanish Language of the Royal Spanish Academy" the Spanish verb "brillar" originates from the Italian verb "brillare". So definitely not Latin, but could be also Italian instead of Spanish.

The noun brillo referring to light, bright, etc is Spanish. The Italian brillo word refers to drunkness

 

[DeschutesCore]

sprockkets[/url]":1jdofo20]
Not knocking you or your comment, but your username, wasn't that a code name of an old Pentium 2 or 3? Interesting.

Yes, Deschutes was the codename for the PII, named after the river in Oregon

I started working in software development right as the PII was coming out, and with Micron and an Intel facility located locally I was fascinated with the leaps the Pentium Pro / Pentium II architecture made over the original Pentium.

I have had the same email address / username ever since.

 

[Dcporter]

Android, Android Wear, Android TV, Android Auto, and Android... Home?

What are they going to call their anthropomorphic robot software?

 

[Deleted member 192806]

Dcporter[/url]":3ay55b47]

Android, Android Wear, Android TV, Android Auto, and Android... Home?

What are they going to call their anthropomorphic robot software?

John.

 

[svgpithon]

My initial thought is that a collaborative team with talent from Android and Nest would handle this, and it seems to me like this would basically be the Android@Home team. Either way, exciting!

 

[svgpithon]

Trandyr[/url]":s390yqpu]

Ali2000[/url]":s390yqpu]"Such devices need to boot up, use an SoC, handle input and output, and communicate over a network—all things the Linux-based Android OS is great at;"

That's such claptrap. You don't need an SoC for most of these IoT things; a microcontroller will work just as well and Android is a bloated OS if you're talking about " light bulbs, door locks, sensors, and whatever other crazy connected objects the IoT crowd dreams up".

Lets put this in perspective. The GE Smart Light Build uses 0.4W on standby (measured by someone on a forum). Not a lot you may think but that is about 50% of what it uses over it's life time. One of the large components of residential electrical demand are vampire loads, devices that are on all the time sucking power. SoCs and Android are the wrong solution.

Then what's the right solution? Honest question here. I'm curious what your take on it is.

I imagine we could keep chugging with specifically designed products based on microcontrollers, but the drawbacks to that are fairly well-known, and clearly make the generic platform approach superior in certain circumstances, like those in which all of these things need to communicate, where making everything "just another Linux machine" really does seem like a good solution. Software developers are freed from handling communications and networking to build features, the way Samsung's doing weird (if controversial) things on top of Android.

 

[cagliano]

"Brillo" in italian means also "drunk"..

quite a bad choice for an os.

 

[UnnDunn]

sprockkets[/url]":37qoeilv]

UnnDunn[/url]":37qoeilv]Great. So now Google will know what all of your IoT things are doing, and be able to target ads at you based on that.

Refrigerator nearly empty? You get tons of ads for grocery stores. Using your "smart" blowdrier? Hello shampoo ads.

Why an ad for a grocery store when the proper response is you are low on xyz?

Why would I need shampoo if I just showered?

People who come up with these stupid scenarios have no clue on how Google's advertising works.

Because <grocery store> paid more to target your demographic than <product xyz> did.

And oh look, you've been using your blowdrier a lot lately, so you're probably running out of shampoo. Let's bombard you with ads for <giant corporate shampoo brand>.

 

[Chemical Tribe]

UnnDunn[/url]":1crhr4i1]

sprockkets[/url]":1crhr4i1]

UnnDunn[/url]":1crhr4i1]Great. So now Google will know what all of your IoT things are doing, and be able to target ads at you based on that.

Refrigerator nearly empty? You get tons of ads for grocery stores. Using your "smart" blowdrier? Hello shampoo ads.

Why an ad for a grocery store when the proper response is you are low on xyz?

Why would I need shampoo if I just showered?

People who come up with these stupid scenarios have no clue on how Google's advertising works.

Because <grocery store> paid more to target your demographic than <product xyz> did.

And oh look, you've been using your blowdrier a lot lately, so you're probably running out of shampoo. Let's bombard you with ads for <giant corporate shampoo brand>.

And what is that grocery going to show ads for?

Using your blow drier is irrelevant and means nothing as far as needing shampoo. That's as accurate as saying I used 100 gallons of water today so I must have done laundry, so let's show ads for that. Or, I flushed my toliet, show ads for toilet paper.

Again, you are showing a lack of understanding on how google ads work.

 

[anoncow]

bthylafh[/url]":1cmjuqy3]
I suspect Google is trying to lower the bar to app developers. It'd be easier for an Android dev to write a Brillo app (assuming its native language is Java-ish or Go, or something else high-level) than an assembly program like I think the devices you're talking about would run.

Even so there are much lighter weight solutions - for example [1] - a full app platform , apps can be written in java or c/c++ , runs on micro controllers and offers a wide variety of IOT/GUI/etc building blocks.

[1]http://www.is2t.com/products/

 

[UnnDunn]

sprockkets[/url]":3ogecr6g]

UnnDunn[/url]":3ogecr6g]

sprockkets[/url]":3ogecr6g]

UnnDunn[/url]":3ogecr6g]Great. So now Google will know what all of your IoT things are doing, and be able to target ads at you based on that.

Refrigerator nearly empty? You get tons of ads for grocery stores. Using your "smart" blowdrier? Hello shampoo ads.

Why an ad for a grocery store when the proper response is you are low on xyz?

Why would I need shampoo if I just showered?

People who come up with these stupid scenarios have no clue on how Google's advertising works.

Because <grocery store> paid more to target your demographic than <product xyz> did.

And oh look, you've been using your blowdrier a lot lately, so you're probably running out of shampoo. Let's bombard you with ads for <giant corporate shampoo brand>.

And what is that grocery going to show ads for?

Using your blow drier is irrelevant and means nothing as far as needing shampoo. That's as accurate as saying I used 100 gallons of water today so I must have done laundry, so let's show ads for that. Or, I flushed my toliet, show ads for toilet paper.

Again, you are showing a lack of understanding on how google ads work.

You seem to think they select ads for you based on what the ad is selling. Nope. They select you for the ad based on the profile you fit. The grocery store says "show our ad to people whose refrigerators are nearly empty", and the shampoo brand says "show our ad to people who use their blow-drier a lot". Since you fit those profiles, you get those ads.

 

[gbjbaanb]

siddharthvader[/url]":17n98r0m]

aimed at ultra low-power devices with as little as 64 or 32MB of RAM.

This made me smile, as the devices I work on have ~256KB of RAM.

Such devices need to boot up, use an SoC, handle input and output, and communicate over a network

You can this kind of thing in ~100KB RAM. A full blown linux based solution seems like overkill - a basic scheduler should be good enough. Especially as we are talking about things like light bulbs and door locks.

and we have a Linux-based OS for IoT announced today called LiteOS by Huwei (cue jokes about the Chinese government knowing what you want to buy - but at least they make those things) which weighs in at 10Kb and comes with all the things a toaster would need, like zero-configuration, auto-discovery and hopefully some security too! They say its already in-use in things like street lights.

 

[colin1497]

thomsirveaux[/url]":3hwvanxf]"Google wants to move in and clean up the fragmented mess by offering Brillo for free to OEMs"

I feel like I've heard this one before.

I was coming here to post this.

 

[Madigon]

Is there an ANSI or IEEE standard for IoT communications? I don't think (opinion) that we need so many OSs as have been presented.

 

[Whiner42]

santos-l-halper[/url]":2cyfwcpj]

UnnDunn[/url]":2cyfwcpj]Great. So now Google will know what all of your IoT things are doing, and be able to target ads at you based on that.

Refrigerator nearly empty? You get tons of ads for grocery stores. Using your "smart" blowdrier? Hello shampoo ads.

Face it: this type of interaction is coming whether it is Google providing it or another vendor. If you are fighting against the future, then good luck. It is a losing battle.

I've said it once and I'll say it again: The Year 2000 called and wants its IoT back.

Except for some very specific use cases (automobiles being a primary one), IoT is the purest marketing baloney ever extruded.

 

[lunarworks]

How does Apple's HomeKit platform compare to this?

Isn't that just a communications protocol that you can implement in your own software/hardware?