Wearable computers (by alaric)
One of my too many projects is to make a wearable computer.
Lots of people are interested in making wearables, but nobody's yet come up with one that hits a "sweet spot" of decent functionality along with it being unobtrusive enough to not be a pain.
Well, I'm a nerd, so I'm far happier to put up with obtrusiveness to get my pervasive cognitive-assistance fix... I've been fascinated by pervasive computers since I was a kid; I read about Steve Roberts' recumbent bicycle as a youngster, as well as plenty of fiction about brain implants and the like.
My two interests in wearable computers are the cognitive boost of an always-available computer, rather than needing to fish in my pocket to pull out a phone and fiddle with its tiny user interface - and the opening of new capabilities such as coupling in a GPS, lights, wireless and wired communications, and so on. Smartphones go a long way towards this, but they're limited - they're aimed at the mass market, rather than people like me, so they have a restricted amount of hardware in them, and little opportunity to improve them. And don't get me started on the lame software.
Now, in order to try and improve on where previous attempts have failed, I've been keeping an eye out for advanced technologies that can make my dreams happen.
My first major breakthrough was chorders as a text-entry device. Many wearable computer systems have been based around these, as they allow fast, easy, text entry while walking around; you only need to spare a hand. But everyone knows about them now. They're catching up!
The next breakthrough ARM's decision to license their microcontroller cores. Now even hobbyists who can't do surface-mount soldering of fine-pitched chips can pick up off-the-shelf modules containing low-power high-performance microcomputers! These aren't PC-grade systems that can run UNIX - they're the sort of CPUs found in low-end smartphones, but still miles ahead of the 8-bit micros Steve Roberts built his bike around, and more than adequate for my needs (while being gratifyingly small, easy to interface to, and with tiny power consumption). A lot of current wearable folks seem to be trying to fit entire PCs onto them - even based on modern netbooks, they're still bulky, power-hungry, and inflexible. Simple is best!
Another awesome advance has been the recent progress in semiconductor light sources. One feature I want on my wearable is decent lighting; for some reason, I seem to need to peer into dark places a lot, so I usually carry two torches on my person whenever I'm outside, and have a head-torch I put on when I'm going outside in the dark. But high-power LED lights are now available at decent prices, so I can include them into my head-mounted display unit, to have a headtorch that's always with me. So while I'm at it, why not have both white and red LEDs, so I can have a dark-adaption-preserving mode as well? Also, some experiments with UV LEDs have shown me that having a UV torch can be pretty handy for finding things in the dark - because a surprising number of things fluoresce in UV. Besides, there's sometimes interesting things written on stuff with UV pens or phosphor-dots. And then there's lasers. Lasers are awesome for pointing at things - I carry a laser pointer, and use it to great effect when the tiny coin cells within it aren't flat. It can point out things on whiteboards, or inside computers, and it can entertain kittens. These days, 200mW laser diodes are easily obtained; they're a bit extreme for simple pointing (so you want a basic 5mW unit for that), but they're great for pointing out things in daylight, communications (you can flash Morse messages to people miles away at night!), and at a pinch, emergency fire lighting. Carrying torches and laser pointers and so on is a reasonable substitute... but it means lots of different things to keep track of, and different batteries that need replacing, and there's no scope for software control to make them automatically flash messages in Morse, or encoding high-speed digital data down a communications laser to receiver towers miles away. And I like to appreciate the finer things in life.
But I still had an issue with packaging. I sketched prototype after prototype, and always struggled with the issues of how to wrap a rugged, waterproof, case. I could design an awesome system as bare PCBs... but they hurt if you fall onto them. Off-the-shelf electronic project boxes are available in a wide range of shapes and sizes, but they're all the wrong shape or size, or they're just not strong enough. I wanted to be able to make the kind of cases mobile phones have - sleek, curved, and basically barely larger than the components within. This was solved for me when I came across an article (I've since lost the URL, sorry!) about a guy who made himself an LED binary watch - and then cast it into a block of clear epoxy resin. The result is incredibly durable - it's not a plastic box around the electronics; the electronics are embedded in the middle of a solid block of plastic. The clearness means that the large display components can be viewed, despite being embedded. Sockets for connections, space for removable parts such as batteries and SD cards, and pushbuttons can be protected during casting by filling them with modelling clay or similar, and pulling it out afterwards. Or you can go for input senses that work through the plastic, such as an embedded camera that spots fingers touching the surface through total internal reflection, cap sense, or using Hall sensors to detect the motion of a magnetic stylus.
So i set about obtaining myself some of the required materials and did some initial experiments to prove the process. That was a success, so for stage 2 I built a test circuit to check for the consequences of shrinkage on electronic components: four LEDs, each with their own series resistor, connected with the longest leads I could manage in a variety of orientations, on a large board, to try and get the maximum shrinkage. Any real circuit would be much more compact than this.
Thankfully, everything worked fine. And the best thing was, the result is totally waterproof. In our highly scientific tests, we ran the LEDs in a sink full of hot water, jumped up and down on them, and ran a car over them. There's some minor surface scratching - but that's it.
For the higher-power parts of my designs (such as that communications laser), I'll need to be careful with heat dissipation; however, I'm already fairly adept at machining aluminium, so some parts of the device might need an aluminium chassis that protrudes out of the resin into a heat sink. Fair enough.
But hard epoxy isn't quite the thing for parts that have to come into contact with me. For that, there's now Sugru. It comes supplied as a clay-like material, but once left to cure for a day, it turns into a tough silicone rubber. It's sticky enough to use as a glue, too; you can just mold pieces of it onto something and let it cure to create a handle. This is great for making ergonomic handles, or other such human-touching parts.
And then, the final piece in the jigsaw, is the Lithium-Ion cell. This little thing packs a lot of stored electrical energy (well, it's stored as chemical energy, but you get the idea), and it's rechargeable. As an added bonus, if you bypass the protection device, they can also be made to explode. In the 90s I turned my nose up at the idea presented in Star Trek that the power cells in phasers could be turned into bombs by breaking some safety device... I ate my words when I learnt about lithium-ion cells. They're easy to charge; you just need to be careful to manage their charging and discharging so as not to trip the safety device and render them useless. So microprocessor control is essential.
And so, finally, it's starting to become practical to build what I want.
My plan is this: I'll have a central module, built into a standard electronic project case so it can be opened and worked on. Inside will be the main lithium-ion battery, plus a reasonably powerful ARM microcontroller as the master control unit, along with an internal SD card slot for mass storage, and charge/discharge control and monitoring logic for the battery, driven by the micro; there'll be sockets for 12v charging power, and USB presenting a serial port for uplinking to a host PC (and reprogramming the micro, if the correct jumpers are set on the board). An SPI bus and power connections from the micro are then exposed to a number of pluggable modules on separate PCBs inside.
The modules will generally be little AVR microcontrollers acting as SPI slaves, driving peripheral devices. I plan to incorporate a GPS receiver, accelerometer, altimeter, gyro and Hall-effect compass as a navigation module, a UART with RS232 line drivers so it can operate as a serial console (handy in the datacentre!), and maybe wifi, bluetooth, or GSM interfaces.
But in the short term, the first modules will be interfaces to other parts of the wearable system. For instance, I want a head-mounted display. The vision in my left eye isn't very good - but a grid of 3mm LEDs encased in a block of epoxy, mounted over that eye onto the front of a pair of safety glasses, can display text and symbols one low-resolution character at a time. If I put a small ARM micro in there, it can perform text-to-speech through an earbud (and maybe even play back MP3s). Also, this is the correct place to mount a bunch of LEDs to work as a head-torch. All of this can be encased in epoxy, with a protruding socket for the earbud, and a link cable down to the central unit (and, perhaps, a couple of buttons so the LEDs can be controlled easily by hand, as well as via software). The link cable would contain power lines, plus high-speed 3.3v RS232; the corresponding module in the central unit would simply have a UART along with a safety fuse for the power lines, coming to an appropriate socket.
I want a wrist unit, instead of a digital watch. An embedded LCD with a backlight would do nicely, with a small AVR micro inside to control it, all encased in epoxy with a Sugru backing. Four buttons along the top edge and three on the bottom, if sited appropriately (with the help of Sugru!) can make it into a chord keyer if I hold my left wrist in my right hand from underneath. And embedded Hall sensors can allow a magnetic stylus to point at things on the LCD. It can be run from the central unit in the same way as the headset, with a power+RS232 line that runs up my sleeve; but a small internal battery (to keep things simple, just a coin cell, accessed via the rear) would enable it to operate independently as a wristwatch when no control commands come from the central unit. Also, this would be an awesome place to mount a 5mW laser module for pointing purposes.
Those two modules alone would make an awesome wearable; an LCD and chorder for command entry, plus a pervasive head-mounted display for visual and auditory output is enough to do most of what I want. An EyeTap would be awesome, but I can't find out how to get one. A conventional head-mounted display is highly attractive, but still means more headset bulk and higher computer requirements. I'm still unsure whether to include some kind of high-res video display for my right eye... Perhaps for version 2; the wonders of a modular design.
But I have a third subsystem in mind. This instructable of a hiking staff that turns into a chair got me thinking. An aluminium tube is light yet strong - and it can contain a terrifying number of lithium-ion cells for a mighty power reserve. That could put out some serious light, run that 200mW communications laser I kept drooling at, provide heated clothing, or run proper two-metre band amateur radio equipment.
So subsystem three is the optional hiking staff. A two-meter aluminium square tube, with a large battery at the bottom (how large will need to be found experimentally; how much weight is comfortable?). A nice Sugru handle at a good height (with an embedded auxiliary chorder, for chording on the go?). A control box mounted just above that, to house the control electronics (an AVR micro, power transistors for controlling the lights, and a few pushbuttons and LEDs for manual control), with a socket for a power+RS232 cable to the central unit - except this time the power is supplied in the opposite direction, as the interface module in the central unit will contain a voltage converter that drops the high voltage from the long lithium-ion stack down to 12v to feed into the external power input, to run the wearable systems and to charge the central unit's battery. Onto the top can be bolted a block of epoxy resin cast around an aluminium chassis, which carries some high-power white and red LEDs (I anticipate one set of white LEDs that are arranged to cover a 360 degree arc, the red LEDs in a 360 degree arc, and a set of LEDs that shine forwards only), and a vertical 200mW green comms laser (so it can be aimed by bringing the staff horizontal, over my shoulder, and letting me sight along its length at distant recipients - or rest it on something for increased stability). Like the wrist unit, it could run independently of the wearable, using its own user interface, so it's useful in its own right.
As usual, though, my main problem is finding time to build all this fun stuff... Maybe once I've retired :-/
Computing, Electronics, Metalworking | alaric | 
Thu 23rd Dec 2010 6:56 pm
Subscribe
By cakemix, Sat 25th Dec 2010 @ 6:03 pm
i love all this too - and get so excited when i think about combining with my crafting!!!