I've been thinking a lot about developing a brake light for a bicycle. The need for such a device has always been obvious to me, but for those of you who may think otherwise, let me explain. Cyclists communicate with drivers and other road users primarily using hand signals. So if you are stopping, you have to let go with one hand and signal. However, if you are doing an emergency stop, it is both difficult and unsafe to signal - both hands need to be on the handlebars to control the bike during an emergency stop. Unfortunately, that's also the most important time when those behind you need to know you're stopping. The situation is essentially a catch-22: if you need to stop quickly, then you should signal to those behind you; but if you signal to those behind you, you won't be able to stop quickly.
In my mind, a brake light is the most effective way for cyclists to signal to other road users. But this isn't a new idea, and this certainly won't be the first to come to fruition. So why don't we see any bikes with brake lights now? From the bike lights I have seen online, there seem to be a few problems with current bike brake lights:
Can't properly sense braking
This mainly applies to those brake lights that use accelerometers to detect the braking. Accelerometers tend to misinterpret bumps and other vibrations as brake applications, causing erroneous lights.
Incompatible with certain types of bikes
Other brake lights use a separate mechanical switch that connects to the caliper or the cable. These may not fit all bikes. Also, they are susceptible to all the wires and the wear and tear of a separate mechanical switch.
Not bright enough
Some of the bike brake lights have only 1-2 small, low power LEDs. What is the use of a brake light if no one can see it?
Concern that other users won't understand the signal
This is the only issue that cannot be solved directly with the device. Perhaps some experiments can be done to test brake lights on unsuspecting motorists/cyclists and see if they understand what the signal means in time to brake. This could end badly...
I will attempt to create a bicycle brake light that solves these problems. Right now I am leaning towards an improved accelerometer sensing brake, that may incorporate another sensor to improve the accuracy of identifying braking periods. But I will also investigate other methods for activating the brake light. Who knows; maybe I'll find some elegant and unobtrusive way to mount a mechanical switch to trigger the brake.