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 is an autonomous agricultural robot developed as an undergraduate research project at SVNIT, published at the AVES 2021 conference. It fuses GPS and magnetometer-guided field navigation with a Bonnet CNN model that segments crops, weeds, and soil in real time — enabling fully autonomous weed removal without human intervention.

Quickstart

Get AGRIBOT running in Gazebo simulation in under 10 minutes

System Architecture

Understand how navigation, perception, and control fit together

Autonomous Navigation

GPS + IMU sensor fusion, noise filtering, and path-planning

Crop-Weed Classification

Bonnet CNN for real-time pixel-level weed segmentation

​
What AGRIBOT Does

AGRIBOT autonomously traverses farm rows using GPS coordinates and magnetometer heading, while a front-mounted camera continuously streams images into the crop-weed classification model. The robot identifies weeds at pixel level and can trigger a removal mechanism without stopping — making it suitable for large-scale precision agriculture.

Gazebo Simulation

Full farm world with textured crop models and simulated sensors

ROS Packages

Catkin workspace packages: description, control, autonomous drive

Hardware Setup

Nvidia Jetson Nano, NEO-M8N GPS, MPU-9265 IMU, Pi Camera

​
Key Capabilities

1

Model the Robot

A SolidWorks-designed four-wheel skid-steer chassis is exported as URDF/Xacro and loaded into Gazebo with simulated GPS, IMU, magnetometer, and camera plugins.
2

Navigate the Field Autonomously

Enter GPS coordinates for the goal waypoint. The gps_converter node converts lat/lon to local XY, computes heading error, and the PD controller drives the robot along crop rows.
3

Classify Crops and Weeds in Real Time

Camera frames are fed into the Bonnet segmentation model. Each pixel is classified as weed (red), crop (green), or soil (blue) at ~2.5 fps on an NVIDIA GPU.
4

Deploy on Jetson Nano

Transfer the trained weights and ROS nodes to an Nvidia Jetson Nano for on-robot inference and autonomous operation in the field.
AGRIBOT was developed and tested primarily in simulation due to COVID-19 restrictions on field access. The software stack is fully operational in Gazebo and has been validated on real farm images.

​
Research & Citation

AGRIBOT was published as “Design of an Autonomous Agricultural Robot for Real Time Weed Detection using CNN” at the AVES 2021 conference (Springer, 2022). 
@misc{https://doi.org/10.48550/arxiv.2211.12077,
  doi = {10.48550/ARXIV.2211.12077},
  url = {https://arxiv.org/abs/2211.12077},
  author = {Patel, Dhruv and Gandhi, Meet and H., Shankaranarayanan and Darji, Anand D.},
  title = {Design of an Autonomous Agriculture Robot for Real Time Weed Detection using CNN},
  publisher = {arXiv},
  year = {2022},
  copyright = {Creative Commons Attribution 4.0 International}
}

Build docs developers (and LLMs) love

Get started for freeTalk to us