Abstract

The gardening industry is a multi-billion-dollar enterprise, and currently lacks an autonomous system for removing weeds from an arbitrary garden. The Mobile Weed Remover (MWR) project in the Advanced Mechatronics and Robotics Lab at Concordia seeks to fill this void.
In this thesis, a great amount of re-design work has been carried out to improve the design of both mechanical and electrical systems of the second generation of MWR to make it functional. The mathematical models of MWR are built including the “ideal modeling” and “discrete modeling”. The simulation of the built models is carried out in Matlab/Simulink to investigate the dynamics of the robot. The control algorithm is designed including forward-backward position correction and error direction and handling. The designed controller is integrated with improved MWR robots and the real time control system is implemented in the on-board computer of the robots. The designed robot uses a camera in order to identify, seek out, and destroy weeds, which enables it to operate in an arbitrary garden. In order to target the roots of the weed, the MWR requires a positioning system that can guide the rover to the desired position, which should be able to take the cues from the visual system. The research in this thesis seeks to adapt a control algorithm to the MWR and enable it to approach a target.
Simulation was conducted to validate the effectiveness of the developed control algorithm. A simple mono-chrome target was used to represent the dandelion in the real time control tests. The experimental results further verify that the control system was operational and deliver satisfactory results.