This has probably been thought of but here goes. I was dreaming of a charger that affixes to the rear of a bicycle frame and is friction driven from the rear wheel. This charger will run a small light on the rear as well as the front of the bike.
It will have an interface to send power to other charging devices (ie. phones, battery chargers,etc).
What say ye good people?
There's a name for that and people use them, I'm no biker so I can't give more information but it does exist.
It was called a bike generator in my days.
When I was a kid, I had two of those tire-driven ‘dynamoes’, as we call it at that time. The friction it added was quite heavy on the pedals to drive the two headlights, but who cares, my bike had bright headlights. In hindsight, early flashaholism perhaps?
Hi
They still exist but in the old days they where quiet unreliable and crude.
Today they are built in into the wheel-hub, are ten times more expensive
but also much more eficiant and produce much less friction/braking.
AFAIK Shimano and Rohloff still sell them.
Most bikers don’t like them because off the added weight.
But i don’t know if they can support charging devices
Mo
Like the old dynamo for bicycles,
have a look at this or this - maybe you can adapt one.
There must be more efficient ones than these “vintage” ones though.
There are still clip on dynamo led lights too if I remember correctly, I’ve not seen one with a device charger built in though, possibly because of the load induced by the generator.
I’ve heard them referred to as bottle generators or dynamos. They clip onto either the frame by the rear wheel or the fork for the front. Some tires come with dynamo tracks (Schwalbe comes to mind) so the little wheel will generate electricity even in the wet. Downsides of the bottle dynamos have already been mentioned: heavy, lots of drag and not very efficient.
There are now bicycle commuters who swear by dynamo hubs. They are expensive and they usually require a custom built front wheel addding to the expense. They are much more efficient than the bottle models but I don’t think they will put out enough power to drive a XM-L as hard as most single 18650 hosts.
Depending on the voltage it gives I’d just add a regulator/step-up/down circuit and a basic USB charger after that. This way you can use the dynamo to charge your cells and use the cells to power up the light.
PS: I don’t know how efficient it would be but a regular Dc motor can be used as a dynamo if you can add a wheel to its axial. Just to try your circuitry or to make your own dynamo.
PS2: A dynamo should give out AC voltage, so a bare motor needs a bridge rectifier to get DC.
Before getting to deep into costs for building something you should consider the power that is used.
To charge one! 18650 you would need to ride for arround 3 hours producing 5V x 1A = 5W
Loading a phone battery (or using it while driving) will take you nearly the same
- So far sounds good
A well driven XM-L alone takes around 10Watts (3to4V x 3A = 9-12W) and thats without
calculating inefficiencies in the electronics
- You would most likely notice this
Now imagine your well illuminated bike with a tripple XM-L light and on top charging your phone
you most probably will get into sweat quiet fast S)
I remember reading that a world-class-rider can produce 200-300Watt peaks for very short time
and at a constant speed more like 100-120Watts on even road while in a race for some hours.
When you are happily riding your bike you most probably won’t do more than 60Watts
I think thats the main reason why most riders use precharged battery-packs and preffer to carry
the weight rather than producing the power while riding. The next hill is always in front of you and
you will need every yota of power you can produce to get to the top. At least i would 
Mo
I had one of the tire driven dynamos on my bicycle back in the late 1950’s. It worked but was very inefficient and created quite a drag. Not to mention that while climbing a hill at slow speed the light would dim to about the brightness of a birthday cake candle.
I did extensive research into this some years back, on average your producing 80-100 watts while riding, and using 10w to power an led means 10-15% of your riding power is to make light, great if you don’t mind the power loss, but i would rather carry a couple 18650 or 26650 batteries instead, under 200 grams to save a lot of pedal power.
The less efficient generators can use up to 50% of your pedal power to generate that 10-15% pedal to electricity conversion, the better ones use maybe 20%
Didn’t the Rohloff/SON dynohub thing turn out to be one of those bogus April Fools’ internet postings that grew legs and refuses to die?
http://www.totolidou.de/bent/en/2010/04/01/10-gear-hub-dynamo-by-rohloff-und-schmidt-son/
It always bothered me that a lot of treadmills, stationary bikes and other exercise devices has to either be plugged in or loaded with batteries to use, so I refuse to do so as a matter of principle (or ???).
No, the hub drive is a fact, just an expensive one. 
it is I believe quite efficient, but to give youban idea, turn on your car, listen to the engine note, Now turn on everything electrical on the car, the rev’s May or may not drop (engine size etc) if the load can slow down a petrol engine at 100hp for arguments sake, your putting one man power in….
Hmmm… not shure for the combination of gear-hub and dynamo-hub but a quick google search
gave me some Shimanos, several SRAM’s and a few SON’s.
See here : Bike-Components.de
and even a readily available USB-Phone-Charger for them :
Supernova The Plug II Plus Dynamo-Ladeadapter
It seems that there is nothing that doesn’t exist,
there are just a few things nobody knows about
(or at least i didn’t)
Mo
My housmate runs a B&M bottle dynamo. It’s been in service for well over 10 years and can be replaced today for around £25. That drives an IQ Fly front light which whilst not super bright really is quite enough for cycling on most roads.
My brother runs one of these:
Schmidt son 28. It’s producing ~6V 3W and he’s driving a Philips head unit and a B&M rear. The front really does light up the road with a lovely beam pattern. It’s quite a bit of kit, service interval is over 50k KM and this has been tested in the real world by a couple of nutters.
Personally the problem I have with both of them is flicker, even when up to speed I can see it. That said I really do envy the simplicity of an always on system with no need for a charger. The Schmidt could drive a USB charging solution but not whilst driving both the lights as it does. I personally wonder if the solution isn’t just to run the dynamo to trickle charge a cell pack. If the pack was dead then you could still get reduced output but you’d also get a more stable supply.
Thanks for all the input folks!