A Roller Based Robotic Hand with Vision based Recognition for Automated Daikon Radish Post-Harvest Processing

This paper presents a robotic hand capable of handling large and heavy agricultural product such as daikon radish without damaging its surface. It aims to integrate multiple post-harvest functions while being flexible in terms of crop type, size and other physical characteristics effectively reducing the number of separate machinery required thus reducing cost and human labor required. This will make modern post-harvest technologies more accessible to small to medium scale producers as well as producers who cultivate seasonal crops. The proposed robotic hand uses soft rollers, pneumatic actuator and computer vision algorithms to achieve this level of integration. The hand can pickup and rotate the daikon radish safely allowing multiple functions to be performed with a single grasp. These functions including computer vision algorithms were implemented and tested with real daikon radish. The force characteristics of the hand, gripping and rolling performance and accuracy of the computer vision algorithms were then evaluated. The results indicate that the proposed robotic hand can identify, grasp and then process heavy agricultural product such as daikon radish reliably.