BNO085 jump detection - which report is appropriate?

Hello,

I want to use a BNO085 for jump detection. I thought this would be easy to see the # of jumps using the stepCounter report, but obviously if the step counter is not fast enough and regularly misses jumps. I tried the shake detector instead, but this also is far from reliable.

I have to say, that during the jumps the Y-Axis point upwards, but anyway the direction of the sensor did not make a difference in terms of reliability of counting each jump.

Does anyone have an idea, if there might be a better, robust solution?

Thanks

We don’t have a pre-built library for jump detection, though one could be made by using a combination of the codes offered in this and the other examples

Try viewing the raw outputs with the example here https://github.com/sparkfun/SparkFun_BN … adings.ino and then construct some code around the readings seen when performing the jumps…you could then combine that with the activityClassifier, and make a ‘jump’ class, or whatever you’d like.

OK, I hoped to have a logic on the BNO’s ARM chip that could handle this, but then I need to do the calculation on the controller side.

Thank you.

Jumps would be followed by a fall, maybe detect a fall and look for a positive spike in the Z axis before the fall and call that a jump?

You can use rotation vector and linear acceleration.

Use the quaternion to rotate the linear acceleration. After, use the z axis (wiil always be positive when going up and negative when going down. Then, you can use a threshold to define what is and what not is a jump.

so I managed to achieve to count jumps, by setting the config of the shake-detector to shake count = 2 and adjust the min/max jump times to 50/300ms respectively. In case anyone is interested…