Abstract: A system for identifying human motion of a subject may comprise: a sensor platform attached to the subject; an attitude filter that receives sensor data and calculates the attitude and heading of the platform; a feature construction and aggregation unit that receives the sensor data and the attitude and heading to calculate a feature vector for a human motion model, an evaluation unit that receives feature vectors and calculates the probability that an instance of a specific human motion occurred; wherein the feature vector comprises: one or more feature constructions, X, such as: forces along and perpendicular to the gravity field direction; a magnitude of rotations perpendicular to the gravity field direction; an adjusted pressure altitude; one or more feature constructions which are time derivatives dX/dt; and one or more feature constructions which are derivatives V(dX/dt), wherein: V(x)=2*H(x)?1, H(x)=+1 (x>0), 0 (x<0), and x is dX/dt.