This repo contains code for our paper -

• Lebin Liang, Dewei Lu, Dong Li, Zilong Chen, Can Wang and Xinyu Wu*

Autonomous navigation in complex outdoor environments is essential for improving efficiency and reducing labor demands in agricultural applications. We presents an open-source 4-wheel steering and 4-wheel drive (4WS-4WD) agricultural autonomous navigation robot platform designed to address the maneuverability and terrain adaptability limitations of conventional robots. By using an omni-directional kinematic model, the system enhances mobility and terrain adaptability in agricultural environments. The system achieves high-precision localization and orientation feedback by integrating a GNSS-RTK receiver, an onboard IMU, and real-time steering angle encoders. The motion controller based on the omni-directional kinematic model ensures precise path tracking and excellent maneuverability. The system's performance is validated through both simulation and real-world deployment, evaluated using reference trajectories. Experimental results demonstrate the system's robustness and efficiency, achieving high-precision path tracking and reliable performance in real-world agricultural applications. All open-source materials, including CAD drawings, circuit boards, and code, can be accessed at the following repository: https://github.com/LebinLiang/four_ws_wd/tree/main

To run this repository, you will need ROS Noetic with Python 3.8. First, set up a catkin workspace as outlined below.

1. Install ROS Noetic
   Follow the instructions on the official ROS documentation: ROS Introduction.

2. Create a New Workspace
   Open a terminal and execute the following commands:

   mkdir -p ~/catkin_ws/src
   cd ~/catkin_ws/

3. Initialize the Workspace

   cd ~/catkin_ws/src
   catkin init

4. Clone the Repository

   git clone git@github.com:LebinLiang/four_wd_ws.git

5. Build the Workspace

   Compilation might require multiple attempts to resolve message dependencies.

   cd ~/catkin_ws
   catkin build

6. Set Up Environment Variables

   source devel/setup.bash

   (Optional) Automatically source the workspace in every terminal session:

   echo "source ~/catkin_ws/devel/setup.bash" >> ~/.bashrc
   source ~/.bashrc

This repository provides various automated scripts for launching core functionalities, located in the /sh folder.

Available Scripts

1. 0-Close-all-terminals.sh - Closes all open terminals.
2. 1-Start-Simulation.sh - Starts the simulation environment.
3. 2-Start-Vehicle-Functions.sh - Launches vehicle functionalities.
4. 3-Data-Collection.sh - Collects trajectory data.
5. 4-Generate-CSV.sh - Converts trajectory data into CSV format.

1. (Optional) Enable or Disable Gazebo GUI

   Modify the gui argument in the 4ws_car/launch/gazebo.launch file:

     <arg name="gui" default="false"/>

• Set false to disable the Gazebo GUI (default).
• Set true to enable the GUI.

2. Start the Simulation

   Open a new terminal and run

   cd sh/
   sh 1-Start-Simulation.sh
   

3. Script Details for 1-Start-Simulation.sh

   The script will automatically open two new terminal windows and an RViz visualization window:

• Terminal 1: Outputs simulation localization data.
• Terminal 2: Displays the navigation controller menu.
• RViz Window: Used to set waypoints and visualize robot trajectories.

Commands executed by the script:

roslaunch taurus_control simulation.launch    # Launch simulation environment.
roslaunch four_ws_navigation four_ws_navigation.launch # Start navigation controller.

Hardware Prerequisites！！

• A NUC computer connected to:
  • An IMU.
  • GNSS receiver.
  • Chassis embedded controller via serial port.

Steps

1. Start Vehicle Functions

   Open a new terminal and run:

   cd sh/
   sh 2-Start-Vehicle-Functions.sh

2. Script Details for 2-Start Vehicle Functions.sh

The script automatically opens five terminal windows and one RViz visualization window, launching:

• Sensor drivers.
• Localization nodes.
• Navigation controllers.
• MAVLink communication nodes (TODO).
• Serial communication nodes.

Commands executed by the script:

roslaunch robot_launch launch_robot.launch         # Start robot sensor nodes.
roslaunch robot_launch launch_location.launch      # Start localization nodes.
roslaunch four_ws_navigation four_ws_navigation.launch # Start navigation controller.
roslaunch mavlink_ros mavlink.launch               # Start MAVLink communication nodes.
roslaunch simple_robot robot.launch                # Start serial communication nodes.

1. Edit predefined trajectories in the four_ws_navigation.py file located at: src/four_ws_wd/four_ws_navigation/scripts/four_ws_navigation.py Example endpoints (auto-linear interpolation between points):

test_goal = [[0,0],[8,0],[8,-5],[0,-5],[0,-10],[8,-10],[8,-15],[0,-15]]
test1_goal = [[0,0],[10,0],[12,-5],[0,-5],[0,-10],[14,-10],[15,-12.5],[15,-15],[0,-15]]
test2_goal = [[0,0],[5,0],[10,0]]

2. Use the RViz interface to manually select waypoints.

1. Use rosbag for data collection.
2. Open a new terminal and execute:

   cd sh/
   sh 3-Data-Collection.sh

3. Topics recorded in the rosbag file:

• /odom
• /target_odom

  rosbag record -o dataset /odom /target_odom

4. Convert rosbag Data to CSV

Update the filename dataset_xxxxxxx.bag in the script:

rostopic echo -b dataset_xxxxxxx.bag -p /odom > odom_pid.csv
rostopic echo -b dataset_xxxxxxx.bag -p /target_odom > target_odom_pid.csv

Then execute:

cd sh/
sh 4-Generate-CSV.sh

1. Visualize Data Using evo_traj

The collected rosbag files can be directly visualized using evo_traj.

2. Common Commands：

evo_ape tum target_odom.tum odom.tum  -r trans_part -v  --plot --plot_mode xy
evo_traj bag dataset_full.bag --all_topic --ref /target_odom --plot --plot_mode xy --save_as_tum

3. For detailed usage, refer to the official documentation: evo

Velocity Control: /cmd_vel Simulated GPS Coordinates: /fix Odometry Data (origin at the starting point): /odom

1. Velocity Control: /cmd_vel
2. Simulated GPS Coordinates: /fix
3. Odometry Data (origin at the starting point): /odom
4. 4WS Parm: /4ws
5. IMU YAW: /imu/imu_yaw

• https://github.com/Romea/romea_controllers

If you found our work useful, please do cite us!