Visual servoing in 3D to enhance full farming in occluded vegetable sceneries

Abstract

In recent years, there has been extensive research and development in the field of agricultural robots. The AgriPV-Bot is a multipurpose field robot to perform selective cultivation in market gardening. However, the cultivation of mixed vegetables in occluded sceneries where not all features are perceptible from top-down perspective poses a challenge especially in spot-farming applications. Therefore, this paper proposes a sweeping motion approach based on 3D visual servoing to utilize multiple view angles to effectively address this challenge. The approach is evaluated both in a simulation and real-world environment.