ROG-Grasp: Root-Oriented Geometry for Robotic Grasping and Placement
For post-harvest agricultural robotics, this provides a practical geometry-based alternative to vision-language-action policies for orientation-aware manipulation.
ROG-Grasp uses root surface geometry from RGB-D perception to estimate produce orientation for stable grasping and placement, achieving high success rates on tomatoes and onions in cluttered scenarios with faster and more reliable execution than VLA policies.
Orientation-aware manipulation is essential in post-harvest agricultural processing, where produce must be grasped and placed in consistent configurations. This paper presents ROG-Grasp, a geometry-based robotic grasping and placement framework that estimates the produce orientation from root surface geometry using RGB-D perception. A YOLO-based root detector and point cloud plane fitting are used to infer the root normal, enabling stable grasp pose generation and orientation-constrained Cartesian motion planning. Experiments on tomatoes and onions demonstrate high success rates and stable execution time in both isolated and cluttered scenarios. Compared with vision-language-action (VLA) policies, the proposed method achieves more reliable and accurate grasp completion with faster execution. These results highlight the effectiveness of geometry-driven perception for practical orientation-controlled manipulation tasks. A video of our paper is available online https://youtu.be/Ir2UtGODdMo.