Mouth-of-Truth

[TOC]

Options

• #### Option-1: RasperberryPi

Linux-like system can do anything.

• Option-2: Arduino + NUC Computer

Sensors on Arduino:

have the Arduino manage sensors (motion detection, capacitive touch) and send a signal to the NUC when specific conditions are met (e.g., "user detected" or "hands on statue").

NUC Responds: The NUC computer would handle the more complex logic, like querying ChatGPT or controlling the TV and speaker, based on the signals received from the Arduino.

Workflow

• When a user stands in front of the statue, the motion sensor activates and prepares the system to interact.

• When the user places their hands on the statue’s mouth, a response is triggered—this could involve illuminating specific LEDs or playing a sound.

• If no motion is detected for a while, the LED strip emits a slow, calming breathing light to signal the system is in standby mode.

• The user types questions via the keyboard, and the system sends the query to ChatGPT. The answer is displayed on the old TV and read aloud via the speaker.

User Interaction Flow

User Movement	Device	Input	Output
Stand Far Away	"Come and ask as yes/no question"		TV/Speaker
	Breathing Light		RGB LED Strip with PWM control
	For detecting user proximity to the statue	PIR Motion Sensor
Put hand inside	To detect hand placement on the statue’s mouth.	Capacitive Touch Sensor or Pressure Sensor
	Light On		RGB LED Strip with PWM control
Ask Y?N question	Light On		RGB LED Strip with PWM control
	Type question	Keyboard - Logitech K400 Plus	TV
Wait	Light blinking		RGB LED Strip with PWM control
	Read out loud the answer		TV/Speaker
Step away	"Come and ask as yes/no question"		TV/Speaker
	Breathing Light		RGB LED Strip with PWM control
	For detecting user proximity to the statue	PIR Motion Sensor

List of Components:

• Raspberry Pi 4 (with WiFi and Bluetooth)

• PIR Motion Sensor (long range)

  • For detecting user proximity to the statue.

• Capacitive Touch Sensor or Pressure Sensor

  • To detect hand placement on the statue’s mouth.

• RGB LED Strip with PWM control

  • For breathing light effect.

• Logitech K400 Plus Wireless Keyboard

  • Mel recommend.

• HDMI to VGA Adapter

  • To connect the Raspberry Pi to the old TV.

• USB or Bluetooth Speaker

  • For audio output.

• HDMI Cable

  • For connecting the Raspberry Pi to the HDMI to VGA adapter.

• VGA Cable

  • For connecting the adapter to the old TV.

• OpenAI API Key

• Old TV

• Multiple breadboard, wires, resistors

  • For connecting sensors and motherboard
  • Probably need soldering

Software

We could either write python or C++ with library support.

1. A simple I/O terminal command-like GUI interface (by default)

   • For input and output the text on TV screen

2. Sensor Inputs

   • Similar as Arduino, need to know GPIO layout
   • 

3. User Inputs (by default)

4. ChatGPT Request

   1. Please answer my question with yes/no, or pick from the following reasons why chatGPT can't answer those questions with yes/no (bad question). These are the bad question categories:
   2. How would you like ChatGPT to respond?
      1. No information access question
      2. Time limit information question
      3. No sensor question
      4. No right or wrong answer question
      5. Dependent on Real-Time Data
      6. Requiring Personal or Contextual Information About the User
      7. Highly Subjective Questions / Personal Opinions
      8. Exact Predictions
      9. Deeply Personal Issues
      10. Medical or Legal Advice
      11. Sensory Input-Based Question
      12. Questions Involving Human Emotions or Relationships
      13. Interpretation of Art or Literature
      14. Speculative or Theoretical Queries
      15. General Knowledge and Fact Verification

5. Display output(sys i/o)

6. Reading the Response Aloud

   • text-to-speech library
   • Different lib might sound differently, need to try and explore which one is the best

Progress

Week2 (09/01-09/07)

Pi System Ready

git clone https://github.com/RadiumLZhang/Mouth-of-Truth.git

Code Structure

1. main.py – The main application file that handles the user interaction flow.

2. sensors.py – Sensor management, including PIR motion and microphone(probably need to adjust later).

3. chatgpt_interface.py – Manages interaction with the ChatGPT API.

4. output_devices.py – Controls TV display, speaker, and LED light.

5. utils.py – Utility functions.

sudo apt-get install python3-rpi.gpio

HC-SR501 PIR Infrared Motion Sensor

https://www.amazon.com/dp/B07KBWVJMP/?coliid=I3B5R132ZHC8H3&colid=2SPLQP9IFVO3J&psc=1&ref_=list_c_wl_lv_cv_lig_dp_it

Unit Test

1. Simulated Signal ✅

python3 test_pir_sensor.py

2. Real Signal ✅

python3 test_pir_sensor_physical.py

Circuit Connection

HC-SR501 PIR Motion Sensor Connection:

​ • VCC → Connect to Raspberry Pi’s 5V pin.

​ • GND → Connect to Raspberry Pi’s GND pin.

​ • OUT → Connect to Raspberry Pi’s GPIO 17.

Week3 (9/8-9/14)

Microphone voice recognition

sudo apt-get install python3-pyaudio
pip3 install speechrecognition --break-system-packages
pip3 install pyttsx3 --break-system-packages # For text-to-speech (optional)

Week4 (9/15-9/21)

LED light

Voltage:DC4-7V

Communication interface:Single-wire communication

LED Chip:WS2812B

The rpi_ws281x library is designed to control WS2812B LEDs on Raspberry Pi.

sudo pip3 install rpi_ws281x

Meet the problem

Then we will do a virtual enviroment

1. Update Raspberry Pi

sudo apt update
sudo apt upgrade

2. Install System-Wide Dependencies

sudo apt install python3-rpi.gpio python3-pyaudio python3-venv

​ • python3-rpi.gpio: For controlling the GPIO pins.

​ • python3-pyaudio: For handling audio input/output (used for voice recognition).

​ • python3-venv: To create and manage virtual environments.

3. Set Up a Virtual Environment for Project

cd ~/Mouth-of-Truth
python3 -m venv venv
source venv/bin/activate

4. Install Python Libraries for the Project

pip install rpi_ws281x
pip install SpeechRecognition pyttsx3
pip install openai

5. Configure GPIO and PWM Permissions

sudo venv/bin/python your_script.py

6. Run Project

source venv/bin/activate
sudo python3 main.py

Circuit Setup

VCC (5V): Connect to the 5V pin

GND: Connect to a GND pin

DIN (Data Input): Connect to a PWM-capable GPIO pin - (GPIO 18).

ChatGPT Request

Week5 (9/22-9/28)

Resolve Internet Issue

1. Tried GT Visitor, EduRoam, GT Other

https://gatech.service-now.com/home?id=kb_article_view&sysparm_article=KB0026531

https://gatech.service-now.com/home?id=kb_article_view&sysparm_article=KB0026877

2. Reinstalled entire PI System -> the python env broken
3. GPIO permission issue

chmod +x main.py

quick fix, might need to have another solution

4. main user flow test
   • While motioned detected
   • Stop breathing light
   • print GPT Request
   • Microphone Input - only one Aux Input, either spreaker or microphone, we need to have another devices to support both functionalities.
   • Spaker Output: play wav file - prerecored.

Week6 (9/29-10/5)

Voice Files

1. Merge the pull request

2. Install required python env-audio

   pip install pydub simpleaudio

3. Reinstall virtual environment

   sudo apt update
   sudo apt install build-essential libssl-dev libffi-dev python3-dev

   pip install openai
   pip install rpi_ws281x --break-system-package
   pip install SpeechRecognition pyttsx3
   pip3 install RPi.GPIO
   pip install pydub
  

Retest Everything

1. LED Light

VCC (5V): Connect to the 5V pin

GND: Connect to a GND pin

DIN (Data Input): Connect to a PWM-capable GPIO pin - (GPIO 18).

2. Motion Detect

VCC → Connect to Raspberry Pi’s 5V pin.

GND → Connect to Raspberry Pi’s GND pin.

OUT → Connect to Raspberry Pi’s GPIO 17.

sudo apt install pipx

1. With background noise test

Week7 (10/6-10/12)

run script

sudo venv/bin/python main.py

Audio

sudo apt-get install portaudio19-dev
sudo venv/bin/pip install pyaudio

sudo apt-get install flac
sudo apt-get install espeak-ng

OpenAI Key

export OPENAI_API_KEY="your_openai_api_key_here"

sudo venv/bin/pip install dotenv python-dotenv

Week8 (10/13-10/19)

1. Change the UI interface font size