Quad-Spider: The 4-legged bluetooth Spider

Introduction: Sharpening the Axe Before the Swing

In the spirit of the age-old lumberjack analogy, this project wasn’t about rushing to build a robot. Instead, it was a meticulously pre-planned engineering experiment, focusing not only on functionality but on modularity, control theory, and real-time human-robot interaction.

Over the summer, I built Quad-Spider, a Bluetooth-controlled quadruped robot that responds to text-based motion commands from a custom Android application created using MIT App Inventor. This article is a deep dive into how I went from concept to prototype, and how I iterated across hardware limitations, software design, and control stability.

---

🦿 Mechanical and Electrical Framework

Robot Chassis & Motion

• Leg Configuration: 4 legs × 3 DOF = 12 degrees of freedom
• Servos: 12 × SG90 servo motors (PWM-controlled)
• Structure: Lightweight acrylic chassis, designed for symmetry and center-of-mass alignment
• Power Source: 11.1V 2200mAh LiPo Battery regulated via a buck converter to 5–6V for stable servo operation
  Spyder model

Pin Mapping

• Servo Control Pins: Digital Pins 2–13 (1 per servo)
• Bluetooth TX/RX: Pins 10 (RX), 11 (TX) via SoftwareSerial
• Arduino Board: Arduino Uno or Nano

---

📡 Bluetooth Communication Architecture

Module: HC-05 Bluetooth Serial Module

• Baud Rate: 9600
• Voltage Divider: 8kΩ & 12kΩ resistors on RX line for 5V → 3.3V compatibility
• Communication Protocol: UART serial interpreted by Arduino and parsed via custom logic

Command Structure

Simple line-based protocol:

"w 5\n" // Walk forward 5 steps
"s 1\n" // Sit once
"k 3\n" // Shake leg gesture 3 times

Each command consists of a keyword and step count, terminated with \n.

---

🧠 Software Stack on Arduino

Core Libraries

• Servo.h for PWM control
• FlexiTimer2.h for periodic timed control of all 12 servos
• SoftwareSerial.h for Bluetooth handling

Servo Timing & Synchronization

• FlexiTimer2 used at 20ms intervals (50Hz PWM)
• Smooth motion via step-by-step interpolation
• Avoided servo jitter and overloads with staggered updates

Command Parsing Logic

Dropped SerialCommand in favor of memory-efficient manual parsing:

if (BTSerial.available()) {
    char c = BTSerial.read();
    if (c == '\n') {
        parseCommand(buffer);
        buffer = "";
    } else {
        buffer += c;
    }
}

This allows precise one-line-at-a-time parsing with low memory overhead.

📱 The Custom App – Built with MIT App Inventor

Features

• Clean UI with command buttons (Walk, Sit, Wave, etc.)
• Input box for custom step count
• Real-time Bluetooth response display via .ReceiveText
• Handles newline-based protocol and shows only the last line for performance

Workflow

• On button click: send “cmd step\n” string
• Use .BytesAvailableToReceive and .ReceiveText(-1)
• Split at \n, select last item from list
• Display in label or debugger box

🧩 Design Thinking & Challenges

Constraint-Aware Coding

• 2KB SRAM limit on Arduino Uno
• Used lean memory buffers and avoided dynamic allocations
• Pin mapping maximized usage without overlap

Power Management

• Buck converter set to ~6V output
• Isolated logic (Arduino) from servo power to prevent resets

Gesture Sequencing

• Each gesture was defined as a series of angles
• Step interpolation added to each transition
• Reduced jerkiness and improved accuracy

📸 Demo & Results

Visit the GitHub Repo to see:

• Arduino code
• App Inventor .aia file
• Full circuit diagrams
• Servo math logic and step models
• Build pictures and videos

🚀 Conclusion: A Spider With a Brain

This robot isn’t just about crawling—it’s about interpreting instructions, executing gait patterns, and wirelessly interacting with a user. It blends mechanical kinematics, embedded systems, and a mobile interface for real-world human-robot communication.

What began as an idea during summer became a fully functional interactive robot—built from scratch using careful planning and relentless debugging.

Feel free to fork the code, use the hardware map, or even integrate your own AI/ML layer for gesture prediction. The world of robotics is yours to spider-crawl into! 🕷️🤖

🔗 GitHub Repository