ros_pid_ga_tuner — GA-Tuned PID Control for a Differential-Drive Robot in ROS

This package demonstrates the navigation of a differential-drive robot using a PID controller whose parameters are automatically optimized via a Genetic Algorithm (GA). The objective is to combine classical control with intelligent optimization to improve goal-reaching performance in simulation.

---

Overview

ros_pid_ga_tuner provides a complete simulated pipeline for goal-oriented navigation of a differential-drive robot running on ROS Noetic. Key components:

• Custom URDF robot model (differential-drive chassis with colored wheels and caster)
• PID controller for linear and angular velocity control
• Genetic Algorithm tuner to optimize PID gains (Kp, Ki, Kd) for both linear and angular controllers
• TF broadcasting of robot pose
• Trajectory publishing as /robot_path for RViz visualization
• ROS launch file with optional goal input parameters (goal_x, goal_y)

This repository is suitable for education, demonstrations, assignments, and as a foundation for further research.

---

Features / Highlights

• GA-tuned PID control: automatically search for good PID gains to minimize trajectory error and settling time.
• Real-time TF pose broadcasting for frame transformations and visualization.
• Path visualization in RViz via /robot_path topic.
• Modular Python node (pid_ga.py) that is easy to extend and tune.
• Launch-time goal parameters to run experiments with different target positions.

---

Repository structure

ros_pid_ga_tuner/
│
├─ package.xml
├─ CMakeLists.txt
├─ README.md
├─ .gitignore
│
├─ launch/
│  └─ pid_ga_robot.launch
│
├─ urdf/
│  └─ differential_robot.urdf
│
├─ src/
│  └─ pid_ga.py
│
├─ rviz/
│  └─ pid_ga.rviz
│
└─ docs/
   └─ screenshots/
       └─ robot_path_screenshot.png

---

Quick start (run demo)

1. Make sure you have a Catkin workspace and place the package in src/.

2. Build:

cd ~/catkin_ws
catkin_make
source devel/setup.bash

3. Make the Python node executable:

chmod +x src/ros_pid_ga_tuner/src/robot_pid_ga.py

4. Launch the simulation with default goal:

roslaunch ros_pid_ga_tuner pid_ga_robot.launch

5. Launch with custom goal:

roslaunch ros_pid_ga_tuner pid_ga_robot.launch goal_x:=3.0 goal_y:=2.0

6. Open RViz (if not auto-started) and set Fixed Frame = world. Visualize /robot_path, TF frames, and the robot model.

---

Nodes (brief)

robot_pid_ga.py

Main Python node implementing:

• PID controllers for linear and angular motion
• Genetic Algorithm optimizer to tune PID gains
• TF broadcasting of the robot pose
• Publishing of /robot_path (nav_msgs/Path) for visualization

---

GA & PID configuration (examples)

• Goal position: goal_x, goal_y (launch params)
• GA parameters: population size, number of generations, mutation rate
• PID limits: v_max (max linear velocity), w_max (max angular velocity)

Tune these for better stability and performance.

---

Messages / Topics

Publishes:

• /cmd_vel — robot control output
• /robot_path — trajectory for RViz
• TF: world → body

Subscribes:

• /odom (optional if using ideal internal simulation)

No custom messages or services required.

---

Troubleshooting & Notes

• If robot oscillates or overshoots: tune GA population/generations, adjust PID limits, or add anti-windup.
• If GA converges poorly: increase mutation rate, population, or modify fitness function.
• Uses ideal odometry (no sensor noise).

---

Suggested further work

• Integrate real robot odometry & sensors
• Add obstacle avoidance (local planner)
• Extend GA to optimize over multiple goals
• Visualize PID error and GA fitness over generations