NOTE: This hexapod project is still under development and may not be fully functional. So, please wait for an official release is coming soon!

Hexapod Robot Control System

A comprehensive control system for a six-legged robot (hexapod) using BeagleBone AI (or Black), featuring both kernel-space drivers and user-space applications with reinforcement learning capabilities.

Project Overview

This project implements a complete software stack for controlling a 6-legged robot with 18 servo motors (3 per leg), including:

• Kernel-Space Components:

  • Linux kernel driver for servo control (PCA9685 PWM controllers)
  • IMU sensor driver (MPU6050) for orientation sensing
  • Hardware abstraction layer for unified device access

• User-Space Components:

  • Kinematics library for precise leg positioning
  • Gait generation for various walking patterns
  • Calibration system for mechanical offset compensation
  • Interactive control interface

• Reinforcement Learning:

  • TD3 (Twin Delayed Deep Deterministic Policy Gradient) implementation
  • Hardware acceleration via TIDL and OpenCL
  • Simulation-to-real transfer capabilities
  • On-robot inference engine

System Architecture

The system follows a layered architecture with hardware abstraction. For visual reference, see the architecture diagrams in the /docs/diagrams/image directory:

• Deployment Diagram: Hardware and software deployment
• Component Diagram: Hardware component interactions
• Sequence Diagram: Runtime interactions
• State Diagram: Hardware state transitions
• Class Diagram: Software class structure
• Building Process: Build and deployment process

.
├── driver/          # Linux kernel device drivers
├── app/             # User applications and libraries
├── docs/            # Documentation
└── scripts/         # Build and utility scripts

Features

• Multiple gait patterns (tripod, wave, ripple)
• Real-time IMU feedback
• Inverse kinematics for precise leg control
• Hardware abstraction layer
• Comprehensive test suite
• Interactive debugging tools

Prerequisites

• BeagleBone AI (or Black) running Linux 4.14+
• I2C enabled (bus 3)
• 18x servo motors (MG996R recommended)
• 1x MPU6050 IMU sensor
• 2x PCA9685 PWM controllers

Quick Start

1. Build the kernel module:

   ./scripts/build.sh -b driver

2. Install the kernel module on BeagleBone AI:

   ./install.sh

3. Build user-space applications:

   ./scripts/build.sh -b app

4. Run the test utility on BeagleBone AI:

   ./test_servo

Development

See individual README files in subdirectories for detailed development guides:

• Kernel Driver README
• User Space README
• Hardware Overview README
• Cross-compile Environment README
• Documentation

Testing

Individual tests:

./test_servo       # Test servo control
./test_mpu6050     # Test IMU sensor
./test_movement    # Test movements like tripod, ripple, ...
./test_calibration # Test servos offset for each leg
./test_hcsr04      # Test HC-SR04 sensor

Troubleshooting

Sensor Issues

• Zero IMU readings: The MPU6050 enters sleep mode to save power. The driver now automatically wakes it before readings.
• Erratic IMU behavior: Check I2C connections and ensure proper power supply voltage.

Servo Issues

• Servos not moving: Check power supply and ensure driver is properly loaded
• Erratic movement: Verify calibration settings and check for loose connections

Documentation

• Hardware Setup
• Software Architecture
• Quick Start Guide

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

License

This project is licensed under the terms of the GNU General Public License v2.0.

Authors

• StrongFood - I mean "TrongPhuc"

Acknowledgments

• BeagleBoard.org
• Linux Kernel Community