Sep 122007
 

WineM solves the problem of remembering all the information about the wines in a large wine rack, cellar or cabinet, or searching through that data. This smart wine rack makes that information always accessible and updatable.

ThingM will be at Wired’s NextFest conference, showing off a prototype of our WineM product. Come visit us if you’re in LA!

You can even read a press release about WineM.

WineM has a bank of RGB LEDs for every cell a bottle can go in. Each cell has a microcontroller implementing a high-power variant of a SmartLED.
There are 576 total LEDs in the WineM prototype, controlled by 52 microcontrollers, bussed together over an I2C bus.

Here’s a little video showing the cell lights in action.


WineM lights test 1

The electronics in each cell look like (click for more details):

Standard through-hole components were used since space wasn’t an issue.

Each cell also has an RFID antenna to recognize the RFID tags on the wine bottles. This lets you recognize which wine bottle is in which cell. A controller board holds the RFID reader and controls up to 12 cells. Four controllers are linked together to in the WineM prototype to control the entire rack:

 Posted by at 1:16 pm
Jul 032007
 

sketching-head.jpg

The Sketching in Hardware 2 conference was a blast. So many interesting people and ideas. I wish we could have it every few months. Mike has his notes and a good summary of this year’s Sketching.

My talk was on “Smart Interface Components”. It was a generalization of the things I’ve been thinking about with the Smart LED prototypes.

sic-title2.jpg

Slides from the talk: sketching07-tod-smartcomponents.pdf

What are Smart Interface Components? Current interface components, the sensors and actuators that comprise the user interface of the gadgets we use, are dumb. They require specialized domain-specific knowledge to make work correctly, non-trivial processing to use, and in general are a pain. Tiny microcontrollers are becoming cheap enough to embed even at the edges of our hardware designs. A component with local brain can embed some of the domain knowledge and enable a higher level of communication between it and the application processor.

An example presented is a Smart LED. LEDs are dumb. Multi-color LEDs are hard to control. Can we make it better?
sic-rgbled.jpg

Imagine an LED that instead of worrying about PWM and current-limiting resistors, you just give it the HSV or RGB color values via a serial line? “#FFCC22 @ 20% brightness”, you say.

Some prototypes (from the flickr set):
sic-protos.jpg

Work is continuing in making production versions of these smart LEDs.

 Posted by at 11:24 am
Mar 252007
 


LEDs should be smarter. Sure we have flashing LED assemblies and even rudimentary RGB-flashing discrete LEDs. But LEDs themselves are predominately just dumb lights. There’s no real reason for this. Fab processes for microcontrollers and LEDs aren’t that dissimilar. It should be possible to have both in a single LED-like package.

There is a glimmer of this happening, like the “RGB LED Slow Colour Change” LEDs you can get. You can see an example of these being used on the Embarrasingly Easy CaseMod. Unfortunately you can’t change the cycle time or anything else of how these LEDs work. So let’s make our own. These prototype Smart LEDs will necessarily be larger than a production run, but the size is getting close, giving us a feeling for how we might use them.

Continue reading »

 Posted by at 7:13 pm
Feb 282007
 

In my periodic perusal of Warren Ellis’s blog, I see he’s linked to Nicolas Nova’s post about “The Ubiquitous Computing of Today”. Nicolas discusses a great LIFT07 paper, “Yesterday’s tomorrows: notes on ubiquitous computing’s dominant vision” by Genevieve Bell and Paul Dourish (Personal and Ubiquitous Computing, 2006). He and the paper highlight a persistent problem with ubicomp research: the ever-receding horizon of technology.

It’s a truism that the technology we need is “just around the corner”. This “proximal future” described in the paper is comforting because it means we don’t have to focus on the actual implementation issues. Just wait for Moore’s Law and everything’ll be alright. But this is a cop-out. I’ve been guilty of thinking this way and it bugs me. I don’t want promises of a future ubicomp world, I want that world right now.

At ThingM we’re focussing on what’s just possible. A tagline of ours is only half-jokingly “ThingM brings you one minute (+/- 20 seconds) from THE FUTURE!” We’re not about the Ubicomp of Tomorrow but rather the Ubicomp After The Next Commercial Break. Our two most recent technology sketches, LoveM and WineM can be built with today’s technology at a cost of 10x over the equivalent dumb version.

An order of magnitude cost increase for a smart version of a dumb object may seem like a lot, but isn’t when put in the context of product variations. Our most pervasive computing device, the cell phone, already exhibits over a factor of ten in price variation, from the $20 unlocked Nokias from yesteryear on Ebay to the tony $700 Sony Ericcson P990 uberphone. Another ubiquitous piece of technology, the TV, ranges from the $20 B&W portable CRT to the mammoth $10k Plasma HDTV.

The markets most allied with ubicomp, electronic gadgets, exhibit enormous price sensitivity, expanding upward in features and price from a minimally-priced base. Most of us buy in the middle to try to hit the sweet spot of the most features with the highest economy of scale. Those of us with special requirements or desires aim at the high-end. This is one reason, as Gibson puts it, the future is already here, just unevenly distributed.

As technology suffuses more into everyday objects, those objects will exhibit the same price elasticity of gadgets. Many domestic objects already do because of luxury and designer brands. The difference in comfort between a no-name leather easy chair from Target and an Eames lounger from Design Within Reach does not track the 10x difference in cost. The cost of adding intelligence to the DWR chair is the same as the sales tax on it. The objects at the high end of the price scale are the ones likely to first incorporate ubicomp technology, making them a better value compared to their non-intelligent siblings. As economies of scale kick in, these new features will become cost-effective for the entire price range.

Exploring what will be possible in a decade’s time is a useful and inspiring task. But until we have nanoassemblers, if we want to impact the lives of people today, we must discover and utilize the technologies available today that are on the verge of having high economies of scale.

 Posted by at 11:47 am
Feb 142007
 

Another technology sketch from my company ThingM. This time it’s Valentine’s Day-themed, with LoveM, a heart-shaped box of “memory chocolates”.


(revver link)

Abstract:

LoveM is a Technology Sketch of an augmented box of chocolates that displays personal memories on an LCD screen as chocolates are removed from the box. It attempts to evoke joy and surprise through the use of available, inexpensive technology embedded into a familiar object. It investigates what happens when we put technology in a non-utilitarian, non-game context and explores the ideas of introducing personal, intimate content into an otherwise mass-produced product.

It’s also our Valentine’s Day present to you. ;-)

 Posted by at 4:14 pm
Jan 152007
 

At my new company ThingM, Mike and I have completed a technology sketch for WineM, a smart wine rack. Below is a video demonstration and an abstract. A full description can be found on the ThingM site. We periodically create Technology Sketches as a way to explore the ideas we’re thinking about.


(revver link)

Abstract:
WineM is a Technology Sketch of a smart wine rack. It’s designed to locate wines in a wine rack using RFIDs attached to bottles and to display which wines have been located using LED backlights behind the bottles. Collectors (or anyone with a large wine cellar) can use it to search through collections, track the location of specific bottles and manage inventory with a minimum of data entry. Linking bottles to networked databases can provide information that would otherwise be too time consuming or difficult to obtain (for example, the total value of a collection, or all the wine that is ready to drink).

 Posted by at 6:44 pm