# Accelerometer Impact/Shock

I am a complete novice when it comes to electrical engineering; however, I'm interested to know the capability of accelerometers.

In addition to capturing data on distance traveled, etc. - is an accelerometer capable of capturing impact/shock (i.e. if attached to a shoe, could it capture the number of times the shoe kicks a ball for example)? If so, would the accelerometer have to directly be impacted or just in the vicinity (i.e. placed anywhere on the foot it would it capture the impact of a ball hitting the foot anywhere)? Also, would an acceleromter be capable of determining between walking/running and the impact of an object hitting the foot (i.e. a ball)?

Yes. Accelerometers measure acceleration, you would see a spike in acceleration from kicking a ball, even if the detector was on your knee. (I assume you mean 3 axis acceleration). The acceleration profile would be quite different from kicking a ball and just running.

Where I am employed we test our product under shock, and we use accelerometers to verify that the shock profile is consistent. This shock profile is a waveform, sampled at 10kHz or more. We are able to confirm that the shock reaches a maximum of 250g and has a duration of 2ms. This is represented in a waveform with hundreds of samples ever the acceleration profile.

In fact measuring distance traveled is not what the accelerometer measures; rather, through calculation of the acceleration profile the speed and distance of the motion can be inferred.

Aside: I do believe that your standard running pedometer does this. They are able to measure speed quite accurately from just the acceleration profile of each step. They integrate the acceleration to get speed. They assume that speed was zero for that duration that the shoe was not accelerating, when it was on the ground. They can also assume that the speed is zero at the end of the stride, so when the integration of acceleration results in non zero at the end they can subtract an overall correction acceleration. Then they integrate speed to get distance interval. They divide the stride length by the stride interval to get the runner speed.

I tinkered with this with my phone, wrote some software to integrate the accelerometer and it was pretty accurate as long as there was no rotation. Rotation complicates the whole situation as you might decelerate on a different axis than you started with; but this is a different topic.

One interesting item I could detect is if you were holding the phone or if it was set on something. Amazingly, you can not for the life of you hold your phone as still as it is when sitting on something fixed (this reaffirmed my theory that the pedometer could detect being still/planted, and thus correct for short term integration errors by assuming zero velocity when planted).

In short, I think you could easily see a different profile between running and kicking. A running stride profile would be a moderate acceleration with a sharp counter acceleration at the end followed by relative stillness. If something were to be kicked, the sharp acceleration would not be followed by stillness, but rather some non-zero values as the foot is still under the limited motion control of your knee and hip.

Sounds like an interesting project actually :)

Accelerometers measure acceleration, not distance traveled. To get distance traveled, you have to integrate acceleration to get velocity and then you have to integrate velocity to get position/distance. This becomes inertial navigation, which is generally only accurate for short intervals before it needs to be reset against a different reference (e.g. GPS).

However, there are many interesting things that you can do with data from accelerometers. It is possible to detect orientation with some accuracy (gravity is an acceleration, so you can easily see which way is down, presuming there are no other accelerations). You can detect taps and bumps or check for freefall conditions.

It is not necessary for the accelerometer itself to get whacked - it just needs to be firmly attached in such a way that it can detect the acceleration of the object in question. If you stock an accelerometer on your shoe, then you can easily detect footsteps and kicks. I would imagine that you could differentiate a step from a kick by looking at the direction of acceleration.

• Ooo, I forgot free-fall – Jon Feb 15 '15 at 7:18

Accelerometers don't measure distance, only acceleration. Think of your car as the accelerometer and you, inside, are a tiny weight. When your car accelerates forward, you feel "pushed back" into the seat and if you slam on the brakes, you are thrown forward. Being pushed back or thrown forward doesn't tell you, directly, how fast you are going, only which way the thing you're in is accelerating. You have to integrate the acceleration values into a velocity value.

Velocity = Acceleration * elapsedTime
Position = Velocity * elapsedTime


-When you pull your foot back to kick the ball, an accelerometer attached to your foot would sense primarily forward acceleration. (+Y)
-When you start to kick, it senses primarily backward acceleration (-Y).
-When your foot contacts the ball, the accelerometer senses abrupt forward acceleration again (+Y).
-Without [other] external forces, accelerometers sense gravity. (-Z)
-While walking/running, the accelerometer would sense forward and backward acceleration, but also the "thump" each time your foot hits the ground.
-While falling, accelerometers measure Z = ~0