Abstract by Nicole Lant
Predicting Body Motion
The Department of Defense is interested in constructing an external device that would assist soldiers when walking or running or lifting heavy loads. To be effective the device would need to anticipate the soldier’s next move. We were given data that monitored movement along three axes at six different sites while the subjects stepped off a platform and over an object. Our first task was to build a model to estimate when a heel lift at the beginning of a step would occur. We produced derivative plots of movement in all axes directions. We found right hip and thigh movement seemed to anticipate left heel lift. However, the instruments were collecting the data at such a small time interval that it was difficult to produce reasonable plots. After thinning the data we were able to accurately locate first and second derivative maximums. We have a preliminary model that predicts heel lift with reasonable accuracy using only two monitors both along a single axis.