3 thoughts on “Mitochondrion Mark 5.1 – Motion sensor data

  1. The traditional solution would be a Kalman filter, and/or a low pass ordinary signal-processingy filter.

    Me, I’d go for a modelling approach that understands that you’re dealing with a spinning staff, not a random abstract thing. Worked for my old firm and its aquatic moving objects.

    1. That’s actually the true orientation after sensor fusion (using Mahony not Kalman filters, so I’m told, I dunno, I didn’t write that bit of code coz maths).

      The sensor and Teensy were moving that fast. Also, it turns out that a Teensy on the end of a USB cable makes a passable poi.

      But yeah, challenge is getting meaning out of the data. Current plan is to simply track the rate of change of yaw to pick times when the staff is vertical and start patterns from that time. If I can spare the cycles to check the motion sensor often enough, then the biggest error will probably be the poor resolution of when patterns can start (~5 ms or so).

Leave a Reply

Your e-mail address will not be published. Required fields are marked *