In this thesis, a machine
vision system based on recent advancements in deep-learning for autonomous harvesting robots
is developed. This thesis firstly reviews recent development in visual sensors, the state-of-the-art techniques
in visual processing and vision processing algorithms based on the deep-learning.