Sensing Sensibly

I recently joined an Iaido club to learn how to wield a Japanese long sword, or Katana. My first challenge was to learn to accelerate the sword quickly, and then stop it before the blade hit the floor or a limb. For some reason, this didn't come naturally, and I quickly formulated a plan to mount an accelerometer on the sensei's sword and one on mine to compare the acceleration patterns. Before I could put the plan into action, I discovered that I was going about it all wrong.

The sensei could tell that I was gripping the sword too tightly at the start, and therefore slowing acceleration of the sword — even when he was observing someone else. So, clearly, he was cueing on something other than sight. On questioning the sensei, it turns out that he could tell the acceleration pattern by sound alone, and from experience knew that improper grip was the most common culprit. 

As shown in the photo, a bokken or wooden sword for Iaido practice has grooves on either side of the blade to replicate the groove on a metal blade. The blade itself makes a noise, and each of the two grooves make a noise as well — that is, if the sword blade is oriented properly during a cut.

So, instead of taking the time to wire an accelerometer to an Arduino and compare patterns, I simply recorded his sword making a downward cut and compared the sound to mine. I used a common audio editing program that ships with the Mac to view the spectral distribution and then — using the sensei's pattern as a template — worked with my sword until the sound patterns generally matched. It'll probably take many months before the frequency distribution patterns are indistinguishable, but at least I have a quantitative goal to reach.

The point of all this is that you shouldn't box yourself into a particular sensor solution. My problem was measuring acceleration and so, naturally, I first thought accelerometer. However, the best — that is easiest, quickest, least intrusive, and cheapest — solution turned out to be a simple digital recorder and common sound analysis software. In retrospect, an accelerometer would have added mass to the sword, the wires would have possibly become tangled, and the balance of the sword would have been affected.

As a practical example, have you ever noticed how the sound changes as you fill a container with water from the tap? As the volume of air diminishes, the frequency of the noise increases. I'm sure that instead of a direct measurement device, you could create a sound-activated switch that would sense a nearly filled container. Another approach would be to use a pair of wires and sense the resistance change. Or, perhaps use an LED and phototransistor to detect the fluid. Or, perhaps a scale to determine when the glass is nearly full of fluid. I suppose you could also bounce an IR or US signal off the surface of the fluid if the glass or other container was large enough.

So, next time you set out to use one of those new inexpensive miniature sensors that are all the rage today, think about your options. They're probably a lot greater than you think. SV