Gesture-Based HMI Controller "Glove MK2"

The device translates hand gestures and orientation into control signals for robotic platforms (UART/Radio) or PC interfaces.

🧠 System Architecture

The firmware is built on a modular "Activity" pattern, allowing the device to switch between different functional roles on the fly.

1. Data Acquisition (Low-Level)

   • Multi-Channel ADC with DMA: 5 bend sensors (flex sensors) are polled in the background. DMA1_Channel1 transfers raw data directly to memory, ensuring zero CPU load for sensor sampling.
   • I2C IMU Fusion: Uses MPU9250 (Accelerometer + Gyroscope + Magnetometer).
   • Complementary Filter: Implemented a real-time sensor fusion algorithm to combine high-frequency gyroscope data with stable accelerometer gravity vectors.
     

     $Angle = 0.96 \cdot (Angle + Gyro \cdot dt) + 0.04 \cdot Accel$

2. Operational Modes (Activities)

   • Car Controller Mode: Translates Pitch (tilt up/down) and Roll (tilt left/right) into a specialized steering/speed packet. Speed is determined by the number of closed fingers (20% increment per finger).
   • Keyboard/HID Mode: Converts finger "masks" into unique character codes sent via Soft UART.
   • Compass Mode: (In Development) Visualizes heading using an RGB LED gradient based on magnetometer data.

3. Smart Calibration Features an interactive calibration routine:

   • Min/Max Capture: The system records raw ADC values for "Open Hand" and "Fist" states.
   • Dynamic Thresholding: Automatically calculates mid-points for binary finger-state detection, compensating for mechanical wear or different user hand sizes.

📡 Output Protocol (Telemetry)

The glove sends 4-byte frames to ensure compatibility with the Robotic Platform v2.5:

• 0xFE — Start Frame
• Servo_Angle — Calculated from IMU Roll
• Speed_Value — Calculated from FingerMask
• Checksum — XOR of data bytes

🛠 Hardware Features

• Status Indication: RGB LED with color-coded modes (Yellow: Calibrating, Blue: Keyboard, Orange: Car Control).
• Power Efficiency: Leveraging STM32 sleep modes and optimized I2C burst reads (14-byte bursts for IMU data).
• Fault Tolerance: Custom HardFault_Handler with visual SOS-code via RGB LED for easier debugging in the field.