Action Determination Using Vertical Acceleration Thresholds

The patent segments the acceleration signal analysis into two distinct components: vertical oscillation analysis (maximum-minimum difference) and step frequency analysis (zero-crossing count). By dividing the recognition criteria into these separate segments, the system can evaluate both vertical motion intensity and stepping rhythm independently, then combine them for more reliable action type determination. This segmentation allows the system to distinguish between walking with knee bending (high vertical oscillation but regular step frequency) and actual running (high vertical oscillation with different step frequency patterns).

The patent introduces a dual-parameter recognition system that changes from using only vertical oscillation amplitude (maximum-minimum difference) to using both vertical oscillation amplitude and step frequency (zero-crossing count). By adding this additional parameter, the system gains more discriminative power to differentiate between walking and running actions. The recognition logic evaluates both parameters simultaneously, allowing accurate identification even when vertical oscillation alone is ambiguous.

The patent transforms the static threshold-based determination into a dynamic multi-criteria evaluation system. Instead of relying on a single fixed threshold for vertical oscillation, the system dynamically evaluates multiple parameters (maximum acceleration, minimum acceleration, zero-crossing count) and applies conditional logic that adapts to different motion patterns. This dynamic approach maintains operational simplicity through structured decision rules while achieving accurate determination through comprehensive parameter assessment.

The patent introduces an intermediary computational layer that processes raw acceleration data through multiple intermediate calculations before reaching the final action type determination. The system calculates intermediate values such as vertical oscillation amplitude, step frequency, and zero-crossing count, then uses these intermediates to inform the final classification. This intermediary processing layer acts as a mediator between simple measurement and accurate determination, maintaining ease of operation while improving precision.