Vision-Guided Targeted Grasping and Vibration for Robotic Pollination in Controlled Environments

TL;DR

This work tackles autonomous pollination in controlled environments by integrating a vision-guided grasping framework with a physics-based vibration model. A 3D plant skeletonization and multi-view perception pipeline yields obstacle-free 7-DoF stem grasp poses, while a discrete elastic rod model guides vibration parameters to maximize pollen release without damage. End-to-end experiments show a 92.5% main-stem grasp success and strong qualitative/quantitative alignment between simulations and real plant dynamics (with some amplitude discrepancy due to modeling simplifications). The results establish a scalable, model-driven path toward autonomous pollination in greenhouses, with potential to reduce labor costs and regulatory barriers in crop production.

Abstract

Robotic pollination offers a promising alternative to manual labor and bumblebee-assisted methods in controlled agriculture, where wind-driven pollination is absent and regulatory restrictions limit the use of commercial pollinators. In this work, we present and validate a vision-guided robotic framework that uses data from an end-effector mounted RGB-D sensor and combines 3D plant reconstruction, targeted grasp planning, and physics-based vibration modeling to enable precise pollination. First, the plant is reconstructed in 3D and registered to the robot coordinate frame to identify obstacle-free grasp poses along the main stem. Second, a discrete elastic rod model predicts the relationship between actuation parameters and flower dynamics, guiding the selection of optimal pollination strategies. Finally, a manipulator with soft grippers grasps the stem and applies controlled vibrations to induce pollen release. End-to-end experiments demonstrate a 92.5\% main-stem grasping success rate, and simulation-guided optimization of vibration parameters further validates the feasibility of our approach, ensuring that the robot can safely and effectively perform pollination without damaging the flower. To our knowledge, this is the first robotic system to jointly integrate vision-based grasping and vibration modeling for automated precision pollination.

Figures (6)

• Figure 1:
• Figure 2: Simulation workflow using PyDiSMech dismechPythonGithub. The plant is modeled as a network of slender rods (skeleton). A moving boundary condition is applied at the grasp point to mimic the vibration actuation, and the resulting dynamics are computed using the Discrete Elastic Rod method.
• Figure 3: The robotic pollination process, illustrated in three key stages: Perception, Reaching, and Shaking. In the Perception stage, the robot arm captures approximately 30 RGB-D images for 3D reconstruction. During the Reaching and Shaking stages, the arm reaches a target grasp position and shakes the plant's stem.
• Figure 4: Generalizability of our skeletonization algorithm, visualized by overlaying the skeletons on the 3D models of 10 diverse plants. The success across all plants using a single parameter set underscores the method's robustness.
• Figure 5: Comparison of experimental and simulated flower motion for Tomato (Better Boy) and Pepper (Majestic Red). (a, b) Flower oscillation amplitude as a function of applied vibration amplitude, showing a clear positive correlation. (c, d) Flower oscillation amplitude as a function of grasping location (distance from root), showing that amplitude decreases as the grasp point moves closer to the flower.