Plucking Motions for Tea Harvesting Robots Using Probabilistic Movement Primitives

This study proposes a harvesting robot capable of plucking tea leaves without cutting them with blades. To harvest high-quality tea, it is necessary to reproduce the plucking motion of breaking the petiole of the leaf, that is, the complicated human hand motion of pulling while rotating. Furthermore, the movement range and the time length of the plucking motion vary greatly depending on conditions that include the maturity of the leaves, thickness of the petioles and length of the branches. In this study, the condition is judged in terms of the magnitude of the reaction force received from the branches. This force is defined as the force from the branches per unit length, when the gripped leaf is pulled up slightly. Using probabilistic learning and a combination of the plucking motions taught directly by a human, new motions are generated according to the force. Thus, the proposed method can realize the plucking motions like that of a human hand. It can generate the range of movement and time length of the motions according to the situations with fewer instructions. Through the generated motions, high-quality tea can be harvested. The effectiveness of the proposed method is verified by evaluating the degree of its similarity with human-taught movements and by conducting actual robot experiments.