An Affordable Navigation and Weed Detection System for Farming Robot

Abstract

Organic farms are characterized with relatively small budgets and there is an existing lack of technology to allow organic farmers to scale up. In this project, an affordable weed detection system and stereo-camera based navigation system is proposed. The crop row detection abilities of the developed navigation system are proved to be extremely robust and invariant to camera orientation. Navigation was tested in a simulation environment for rapid development and a optimal controller was implemented for the real robot. By calculating the time it takes to remove weeds via a shortest path algorithm, it was proved that a farming robot could drastically cut down labor costs while increasing profit and increasing the area weed removal may be performed on.