Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/Dhruv2012/Autonomous-Farm-Robot/llms.txt

Use this file to discover all available pages before exploring further.

AGRIBOT requires a full ROS Melodic installation with Gazebo 9 and several sensor simulation plugins to run the autonomous navigation stack. On top of the ROS layer, the CNN crop-weed classification pipeline depends on a set of Python packages built around TensorFlow 1.x and Keras. This guide walks through every installation step in order, from the base OS packages to building the catkin workspace.

​
System Requirements

Before installing, confirm your environment meets these minimum specifications:

Operating System

Ubuntu 18.04 LTS with ROS Melodic (recommended), or Ubuntu 16.04 with ROS Kinetic

Memory

At least 8 GB RAM. An NVIDIA GPU is recommended for real-time Bonnet CNN inference (tested on NVIDIA 940 MX, avg. 2.5 fps)

Python

Python 3.6 or later (required by the autonomous_drive and classification packages)

Simulator

Gazebo 9 or later (included with ros-melodic-desktop-full)

​
ROS and Gazebo Installation

Install ROS Melodic from the official ROS package repository. The desktop-full variant includes Gazebo 9, RViz, and all standard ROS tools in a single install. 
sudo sh -c 'echo "deb http://packages.ros.org/ros/ubuntu bionic main" > /etc/apt/sources.list.d/ros-latest.list'
sudo apt-key adv --keyserver 'hkp://keyserver.ubuntu.com:80' --recv-key C1CF6E31E6BADE8868B172B4F42ED6FBAB17C654
sudo apt-get update
sudo apt-get install ros-melodic-desktop-full
echo "source /opt/ros/melodic/setup.bash" >> ~/.bashrc
source ~/.bashrc
After installation, verify that roscore launches without errors before proceeding.

​
ROS Dependencies

AGRIBOT relies on several ROS packages beyond the desktop-full bundle. Install them all with a single apt-get chain: 
sudo apt-get install ros-melodic-hector-gazebo-plugins
sudo apt-get install ros-melodic-gazebo-ros-control
sudo apt-get install ros-melodic-ros-control ros-melodic-ros-controllers
sudo apt-get install ros-melodic-teleop-tools
sudo apt-get install ros-melodic-move-base
sudo apt-get install ros-melodic-mapviz ros-melodic-mapviz-plugins
sudo apt-get install ros-melodic-multires-image
The table below explains what each package provides in the AGRIBOT stack:
PackageRole in AGRIBOT
hector-gazebo-pluginsSimulates GPS (libhector_gazebo_ros_gps.so), IMU (libhector_gazebo_ros_imu.so), and magnetometer (libhector_gazebo_ros_magnetic.so) on the robot model
gazebo-ros-controlBridges Gazebo’s physics engine with ros_control effort controllers
ros-control / ros-controllersProvides PID effort controllers for the four wheel joints
teleop-toolsEnables keyboard and joystick teleoperation for manual testing
move-baseAction server used by AutonomusDriveAction.py for waypoint patrol navigation
mapviz / mapviz-pluginsGPS trajectory visualization alongside the Gazebo simulation
multires-imageRequired by the Mapviz tile-map plugin for satellite imagery overlays

​
Python Dependencies for Classification

The crop-weed classification pipeline uses a Bonnet CNN architecture trained on the CWFID and Bonn sugar beet datasets. Install all Python dependencies with pip3: 
pip3 install tensorflow==1.14 keras numpy matplotlib opencv-python scikit-learn scikit-image Pillow imutils keras-metrics
The classification code in Crop_Weed_Classification/model.py was written for TensorFlow 1.x and Keras 2.x. The pip-installable version that matches the original development environment is tensorflow==1.14. Upgrading to TensorFlow 2.x will introduce breaking API changes — particularly around keras.backend, session handling, and model loading. Test compatibility thoroughly before upgrading your environment.

​
Build the Catkin Workspace

With all system and Python dependencies in place, build the AGRIBOT catkin workspace: 
cd Autonomous-Farm-Robot/agribot_ws
catkin_make
source devel/setup.bash
catkin_make compiles the following packages in dependency order:
  • agribot — top-level meta-package
  • agribot_description — URDF/xacro models, Gazebo worlds, maps, and launch files
  • agribot_control — PID controller configuration and spawner launch file
  • autonomous_drive — GPS conversion, sensor fusion, and autonomous drive Python nodes
Add the workspace source command to your ~/.bashrc so that every new terminal has the AGRIBOT packages on its ROS path automatically:
echo "source ~/Autonomous-Farm-Robot/agribot_ws/devel/setup.bash" >> ~/.bashrc
source ~/.bashrc

​
Verify Installation

Confirm that ROS can locate all three core AGRIBOT packages: 
rospack find agribot_description
rospack find agribot_control
rospack find autonomous_drive
Each command should print the absolute path to the package directory without any errors. If a package is not found, re-run catkin_make in agribot_ws and confirm that the workspace devel/setup.bash has been sourced in the current shell.

Build docs developers (and LLMs) love

Get started for freeTalk to us