Custom Dynamic Window Approach (DWA) Local Planner

Technical Assessment: Robotics Software Apprentice | Origin (Formerly 10xConstruction)

This repository contains a comprehensive, from-scratch implementation of the Dynamic Window Approach (DWA) local planner developed for the ROS 2 Humble framework. This project serves as a technical submission for the Robotics Software Apprentice position at Origin, demonstrating proficiency in autonomous navigation, kinematic modeling, and real-time obstacle avoidance.

---

🚀 Project Overview

The objective of this project is to automate the navigation of a TurtleBot3 Burger within a simulated environment. By implementing the DWA algorithm independently of existing libraries such as Nav2 or DWB, this project showcases a fundamental understanding of local path planning and motion control.

Core Functionalities

• Independent Algorithm Implementation: The DWA logic was developed entirely in Python, adhering strictly to the requirement of avoiding pre-existing navigation plugins.
• Goal-Oriented Navigation: The system successfully navigates to specified coordinates with high precision and automated termination upon reaching the goal tolerance.
• Collision Avoidance: Real-time LiDAR data from the /scan topic is utilized to ensure safe passage through dynamic and static obstacles.
• Enhanced Visualization: The planner provides telemetry and trajectory candidates via RViz MarkerArray messages for comprehensive system auditing.

---

🛠 System Configuration and Installation

This project is deployed within a VS Code Dev Container to ensure a standardized and reproducible development environment, mitigating cross-platform dependency conflicts.

1. Environment Initialization

• Clone this repository.
• Open the directory in Visual Studio Code.
• Select the "Reopen in Container" option when prompted by the editor.

2. Dependency Management

Execute the following commands within the container terminal to install the necessary TurtleBot3 simulation packages:

# Navigate to the workspace source directory
cd ~/turtlebot3_ws/src

# Clone official ROBOTIS dependencies for Humble
git clone -b humble https://github.com/ROBOTIS-GIT/turtlebot3_msgs.git
git clone -b humble https://github.com/ROBOTIS-GIT/turtlebot3.git

# Compile the workspace
cd ~/turtlebot3_ws
colcon build --symlink-install
source install/setup.bash

3. Environment Specification

Define the TurtleBot3 hardware model for the simulation:

export TURTLEBOT3_MODEL=burger

---

🏃 Simulation Execution

A unified launch file is provided to initialize the Gazebo environment, spawn the robot model, and execute the custom DWA planner node simultaneously.

ros2 launch dwa_planner_py dwa_sim.launch.py

Goal Specification

A navigation goal may be assigned via the RViz "2D Goal Pose" interface or by publishing directly to the /goal_pose topic:

ros2 topic pub /goal_pose geometry_msgs/PoseStamped "
header:
  frame_id: 'odom'
pose:
  position: {x: 3.0, y: 0.0, z: 0.0}
  orientation: {w: 1.0}"

Visualization and Analysis

To inspect the internal state of the planner:

1. Initialize RViz: rviz2
2. Configure the Fixed Frame to odom.
3. Add a MarkerArray display subscribed to the /dwa_trajectories topic.

Visual Indicators:

• Blue Vectors: Represent the set of sampled candidate trajectories.
• Red Vector: Represents the optimal trajectory selected for execution based on the cost function.

---

🧠 Technical Methodology

The DWA planner operates through a four-stage cyclic process:

1. Velocity Sampling: Linear and angular velocities are sampled within the dynamic and kinematic constraints of the robot.

2. Trajectory Prediction: The system predicts forward states for a 2.0-second horizon using a discrete-time unicycle kinematic model.

3. Cost Function Evaluation: Each candidate trajectory is scored based on weighted objectives:

   • Target Proximity: Minimizing Euclidean distance to the goal.
   • Safety Margin: Maximizing clearance from LiDAR-detected obstacles.
   • Alignment: Minimizing heading error to ensure smooth, efficient motion.

4. Command Publication: The velocity pair yielding the minimum cost is published to the /cmd_vel controller.

---

📚 References

The development of this planner was informed by the following academic and technical resources:

• Primary Algorithm: The Dynamic Window Approach to Collision Avoidance (Fox, Burgard, and Thrun, 1997)
• Technical Tutorials: Concepts of DWA implementation.
• AI Assistance: Generative AI tools were utilized to assist with documentation formatting and debugging.

---

📂 Repository Structure

dwa_planner_py/
├── launch/             # Orchestration files for simulation and node deployment
├── worlds/             # Gazebo world files containing obstacle configuration
├── dwa_planner_py/     # Primary Python implementation of the DWA node
├── setup.py            # Package metadata and entry point definitions
├── package.xml         # ROS 2 manifest and dependency declarations
└── README.md           # Documentation and setup guide