Hardware: M5Paper. 2. Acquisition: Read raw coordinates from GT911 via I2C. GT911 supports multi-touch. Store contact points in a dynamic array. 3. Processing: Implement smoothing algorithm (e.g., Kalman filter) to reduce touch noise. Calculate displacement vectors between frames. 4. Recognition: Define patterns (swipe, pinch, rotate). Use a finite state machine to track gesture sequences. For example, detect 'pinch' by calculating distance variation between two contact points over multiple iterations. 5. Verification: Test responsiveness on 960x540 screen. Adjust sensitivity thresholds. 6. Deep Dive: Integrate a lightweight machine learning model (TinyML) on the ESP32 to accurately recognize hand-drawn gestures.