Sound-Tracking Bot

📋 Description

Sound-Tracking Bot is an autonomous quadruped robot controlled by acoustic detection and capable of:

• 🎵 Localizing a sound source via stereo cross-correlation
• 🤖 Moving autonomously or via teleoperation
• 🌐 Communicating via WebSocket for remote control
• 🎚️ Filtering and processing audio signals in real-time

The system runs on an ESP32 with stereo I2S audio acquisition and control of 12 servo motors via PCA9685.

---

🏗️ Architecture

Project Structure

src/
├── main.cpp                    # Entry point, FreeRTOS tasks
├── bot/                        # Robot control
│   ├── bot_ctrl.cpp/hpp        # Inverse kinematics, locomotion
│   └── utils/
│       ├── ik_calcul.cpp/hpp   # IK calculation for legs
│       └── pca9685.cpp/hpp     # PWM driver for servos
├── sound/                      # Audio processing
│   ├── i2s_sound_acquisition.cpp/hpp  # Stereo I2S acquisition
│   ├── st_sample_data.hpp      # Audio data structure
│   └── utils/
│       └── biquad_filter.cpp/hpp      # Bandpass filter
└── ws_com/                     # WebSocket communication
    ├── web_socket_server.cpp/hpp      # WebSocket server
    └── st_cmd_data.hpp         # Received commands

---

🎯 Key Features

1️⃣ Acoustic Detection

// Sound source localization via L/R signal cross-correlation
// Returns the angle of the sound source in degrees
int calculate_angle(const std::vector<float>& sigL, const std::vector<float>& sigR)

• Sampling frequency: 44.1 kHz
• Frequency band: 1-1.2 kHz (bandpass filtering)
• Microphone distance: 10 cm
• Resolution: Windows of 441 samples

2️⃣ Inverse Kinematics

std::array<double,3> ik_leg(const std::array<double,3>& target, ...)
// Returns [hip_angle, knee_angle, foot_angle]

• IK calculation for 4 legs (2 front, 2 rear)
• Lengths: coxa=60mm, tibia=76.84mm, tarsus=128.05mm

3️⃣ Locomotion

• Autonomous mode: Sound tracking with tripod gait
• Teleoperated mode: WebSocket control
• Gait parameters: Step length=130mm, height=70mm, period=2s

4️⃣ WebSocket Control

ws://192.168.4.1/ws

Commands:
- "ping" → "pong"
- "get_angle" → angle in degrees
- "set_auto-true/false"
- "set_manual-true/false"
- "set_teleop-true/false"
- "set_qTarget-<float>-<index>"
- "set_qActive-<true/false>-<index>"

---

🔧 Hardware Configuration

ESP32 D1 Mini

Component	GPIO	Notes
I2S Audio		Stereo acquisition
BCK (Bit Clock)	26
WS (Word Select)	25
DATA_IN	17	L/R 32-bit data
PCA9685	I2C	PWM servo driver @ 50 Hz
SDA	21
SCL	22
WiFi	SoftAP	SSID: ESP_Spider

Servo Motors

• Count: 12 (3 per leg × 4 legs)
• Range: 0°-180°
• Frequency: 50 Hz

Electrical Schematic

---

🚀 Quick Start

1. Configuration

# Copy configuration
cp sdkconfig.esp32_d1_mini sdkconfig

2. Build and Upload

idf.py build
idf.py flash monitor

3. WebSocket Connection

# Via wscat
wscat -c ws://192.168.4.1/ws

---

📊 Audio Signal Processing

Audio Pipeline

I2S Input (44.1 kHz) 
    ↓
Biquad Filter (1-1.2 kHz)
    ↓
Sliding window (441 samples)
    ↓
L-R cross-correlation
    ↓
Angle calculation (arcsin + degree conversion)

Bandpass Filter

void setup_bandpass(float f1, float f2, float fs)
// Center frequency: sqrt(f1*f2)
// Quality factor: sqrt(f2/f1)

---

🤝 Operating Modes

Autonomous Mode

• Active sound listening
• If valid angle: turns toward sound source
• If angle < 20°: moves forward
• Otherwise: continues turning

Manual Mode

• Direct control of target angles
• Individual servo enable/disable

Teleoperated Mode

• run (move forward) and turn commands
• Customizable turning angle

---

📈 FreeRTOS Tasks

Task	Priority	Period	Function
sound_task	5	Continuous	I2S audio acquisition
cycle_task	4	100 ms	Motor control, processing

---

📚 Références

• ESP-IDF Documentation
• FreeRTOS

---