After I deployed an initial prototype, it occurs to me that actual knowledge of steps (while, yes, walking) would be helpful in creating my mapping strategies (translation of dancer motion to music).
Right now, I am looking at RRACE:
http://www.researchgate.net/publication/257307889_RRACE_Robust_realtime_algorithm_for_cadence_estimation
It estimates cadence, which will be helpful. It introduced me to the Lomb normalized periodogram. It is like the FFT, but for non-uniformly sampled data.
I am also measuring stationary moments in each leg...Using these two tools, I hope to extrapolate with reasonable accuracy when people are stepping. I can then use the distance between steps (literal distance using dead-reckoning, using the algorithm from xIO technologies)... or simply the time between steps... distance might be more interesting?)
No comments:
Post a Comment