Piezoelectric MEMS Contact Detection Using Audio-Motion Correlation

The patent segments the contact detection problem into multiple independent signal processing streams: acoustic signal processing, vibration signal processing, and feature extraction. Each stream handles specific aspects of contact detection separately, then results are integrated for final classification. This segmentation reduces the complexity of processing raw signals directly and improves classification accuracy by allowing specialized processing for each signal type.

The patent introduces intermediate feature extraction and comparison stages between the raw sensor signals and the final classification output. These intermediary processing steps include extracting relevant features from acoustic and vibration signals, comparing them against reference patterns, and integrating results before making contact type determinations. This intermediary processing layer simplifies the overall system by breaking down the complex classification task into manageable stages.

The patent merges multiple signal processing approaches and sensor inputs (acoustic signals from microphones, vibration signals from piezoelectric MEMS) into a unified contact classification system. By combining information from multiple independent signal sources and processing them through integrated feature extraction and comparison stages, the system achieves more reliable contact classification than any single processing method could provide alone.

The patent transforms raw sensor signals into different parameter representations through feature extraction, converting time-domain signals into frequency-domain characteristics and other relevant features. This parameter transformation allows the system to identify contact patterns more reliably by analyzing signals in terms of their characteristic features rather than raw waveforms, effectively filtering out noise while preserving contact information.