Multi-Sensor Gesture Recognition With Adaptive Sampling Cycles

The patent implements dynamic sampling cycle adjustment where the sensor sampling rate is adaptively changed based on the detected movement state. During periods of no or minimal movement, the sampling cycle is extended to reduce power consumption. When movement is detected (indicating potential gesture activity), the sampling cycle is shortened to improve recognition accuracy. This dynamic adjustment resolves the contradiction between precision and energy consumption.

The system employs periodic sampling with variable periods, switching between long sampling cycles (for power saving) and short sampling cycles (for accurate detection). This periodic action with adaptive period length allows the system to maintain gesture recognition capability while significantly reducing average power consumption during idle periods.

The patent segments the sensing function across multiple sensor types, each optimized for specific detection tasks. The acceleration sensor handles large-scale movements, the gyro sensor captures rotational movements, and optional biometric sensors detect fine finger movements. This segmentation allows the system to achieve high recognition accuracy without requiring all sensors to operate at maximum capacity simultaneously, thereby managing complexity more effectively.

The system dynamically changes operational parameters (sampling rates, activation states) of different sensors based on the current context and detected movement patterns. Not all sensors operate at full capacity continuously; instead, their parameters are adjusted to provide necessary data while minimizing power consumption and effective complexity. This parameter optimization resolves the contradiction between accuracy and complexity.