I've been spending ages trying to work out the optimal angle for the placement of a single accelerometer for epileptic seizure recognition.
Everytime I spend ages doing something I suddenly realise that what I've been doing has been flawed and have to try something else... well now I think I've got it... and the diagrams are pretty too!
The above would be a plot of the components of motion with any direction of accelerometer. The further away from the center (and the hotter the colour), the better that angle would be for picking up the motion the graph is for. I'm well pleased with this. Now all I've got to do is make it make lots of these and mark angles where blue happens as not good angles for placement... ho humm...!