Sentry

Overview

The Sentry challenge involves STM32F429ZI's Gyroscope to capture data, process it and then compare the sequence via DTW recognition for gesture-based unlocking/locking.

Demo Video

Link: https://youtube.com/shorts/s6QqmGTkBAY?si=GU7sCO6Fnp7VEv-T Challenge demonstration includes:

1. Gesture recording.
2. Gesture testing and unlock.
3. Deleting gestures.
4. Live Teleplot visualization.

Architecture:

• System pattern flow: Uses queue and button interrupt events for saving/recording gestures.
• Collection: Uses in-build gyro sensor via spi communication interrupts for collecting x, y, z axis data.
• Filtering: Uses Moving Average FIR (Finite Impulse Response) window filtering for processing axis raw data.
• Fetching: Code saves/load a gesture in a file on dev board flash when required to record/match.
• Screens: Uses in-built lcd driver for three types of visual feedback: success, failure and loading.
• Recognition: Uses Dynamic time warping (DTW) algorithm to detect the matching rate of the gestures.

System Design Flow:

• Secret gesture recording: User record a "secret" sequence gesture using a "Record Key."
• New gesture recording: User record a "new" sequence gesture using a "Record Key."
• Gesture matching: Algo matches the secret and new sequence gesture to "Unlock" if success and remains "Locked" if does not matches.

---

Features

1. Data Collection:

   • Captures gyroscope data (X, Y, Z) for a gesture sequence.
   • Filters raw data for noise reduction and stability.
   • Displays live data on Teleplot.

2. Gesture Recording:

   • Press the "Record Key" to record a gesture.
   • Recorded data is stored in a file on the microcontroller (gesture_data.txt).

3. Gesture Matching:

   • Press the "Enter Key" to test a gesture.
   • Compares the test gesture to the saved gesture with configurable tolerances.

4. Visual Indicators:

   • Green LED: Successful unlock.
   • Red LED: Unlock failed.

5. File Management:

   • Gesture data is saved to a file.
   • Press the "Delete Key" to remove the saved gesture.

6. Auto Lock:

   • Automatically locks after 10 seconds of inactivity.

---

System Requirements

• Hardware:
  • STM32F429ZI development board with an integrated accelerometer and gyroscope.
  • LEDs for visual feedback.
  • Buttons for recording and entering gestures.
• Software:
  • PlatformIO with the Mbed framework.
  • Teleplot for live data visualization.

---

Setup Instructions

1. Clone this repository:

   git clone https://github.com/VeronicaGupta/Sentry.git
   cd Sentry

2. Install dependencies:

   • Ensure PlatformIO is installed in your VSCode environment.
   • Open the project folder in PlatformIO.

3. Configure the platformio.ini file:

   [env:disco_f429zi]
   platform = ststm32
   board = disco_f429zi
   framework = mbed
   
   build_flags = 
       -D MBED_CONF_PLATFORM_STDIO_CONVERT_NEWLINES=1
       -D MBED_CONF_PLATFORM_MINIMAL_PRINTF_ENABLE_FLOATING_POINT=1
   monitor_speed = 115200

4. Flash the firmware:

   • Connect the STM32 board to your computer.
   • Build and upload the code using the PlatformIO Build and Upload buttons.

5. Launch Teleplot for live data visualization:

   • Configure Teleplot to connect to the serial COM port at 9600 baud.

---

Usage Instructions

1. Recording a Gesture:

   • Press the "Record Key" (e.g., USER_BUTTON) to start recording.
   • Perform the desired gesture while holding the board in your hand.
   • The system records for 3 seconds and saves the gesture.

2. Testing a Gesture:

   • Press the "Enter Key" to test a gesture.
   • Perform the gesture and wait for feedback:
     • Green LED: Unlock successful.
     • Red LED: Unlock failed.

3. Deleting a Gesture:

   • Press the "Delete Key" to delete the saved gesture.

4. Auto Lock:

   • The system locks automatically after 10 seconds of inactivity.

---

Code Structure

src/
│   main.cpp           # Main application code
include/
│   gesture.h          # Gesture processing functions
│   file_io.h          # File handling for gesture data
│   imu.h              # IMU setup and data collection

---

Key Algorithms

1. Gesture Filtering:

   • Implements a low-pass filter to reduce noise and stabilize gyroscope readings.

2. Gesture Matching:

   • Compares the saved gesture to the test gesture using element-wise tolerances.

3. File Handling:

   • Saves and loads gestures from gesture_data.txt.

4. Event Handling:

   • Interrupt-driven design for buttons.

---

Grading Criteria

Criteria	Details	Weight
Objectives Achieved	Functional lock/unlock system with gestures	40%
Repeatability and Robustness	Consistent gesture matching	20%
Ease of Use	Intuitive interaction with the system	10%
Creativity	Innovative design elements	10%
Well Written Code	Clean, modular, and documented code	10%
Complexity	Advanced algorithms or features	10%

---