Crop row finder

Abstract

In an effort to increase the productivity of modern agriculture a host of robotic platforms and machinery has seen the light of day. This thesis explores the pos- sibilities for developing a system that enables a small mobile robot to navigate in crop rows using visual feedback. To increase the reusability of the software developed, it has been implemented as a plug-in for Mobotware. It utilizes OpenCV functions to accomplish the necessary image analysis used for navi- gation. To control the robot a scripting language is used that contains useful premade functionalities that can be readily applied to an array of hardware setups. System verification is based on tests performed using the robot in an emulated field environment.

full thesis

http://aut.elektro.dtu.dk/mobotware/doc/robotti/Robotti_row_follow_Aske_Bay_Jakobsen.pdf