As the end execution tool of agricultural robots, the manipulator directly determines whether the grasping task can be successfully completed. The human hand can adapt to various objects and achieve stable grasping, which is the highest goal for manipulator design and development. Thus, this study combines a multi-sensor fusion tactile glove to simulate manual grasping, explores the mechanism and characteristics of the human hand, and formulates rational grasping plans. According to the shape and size of fruits and vegetables, the grasping gesture library is summarized to facilitate the matching of optimal grasping gestures. By analyzing inter-finger curvature correlations and inter-joint pressure correlations, we investigated the synergistic motion characteristics of the human hand. In addition, the force data were processed by the wavelet transform algorithms and then the thresholds for sliding detection were set to ensure robust grasping. The acceleration law under the interaction with the external environment during grasping was also discussed, including stable movement, accidental collision, and placement of the target position. Finally, according to the analysis and summary of the manual gripping mechanism, the corresponding pre-gripping planning was designed to provide theoretical guidance and ideas for the gripping of robots.
Keywords: agricultural manipulator; grasp analysis; grasp planning; manual simulation; tactile glove.