SG-DOR: Learning Scene Graphs with Direction-Conditioned Occlusion Reasoning for Pepper Plants
2026-03-06 • Robotics
RoboticsComputer Vision and Pattern Recognition
AI summaryⓘ
The authors created a system called SG-DOR to help robots figure out which leaves are blocking a fruit in dense plant areas, like pepper plants. Their method looks at detailed 3D points of plant parts and builds a map showing how parts are connected and which leaves block the fruit from specific directions. They tested their approach on computer-made pepper plant data and showed it can better predict leaf blockage and connections than other methods. This helps robots plan how to pick fruits without damaging the plant.
scene graphocclusion reasoningpoint cloudgraph neural networkinstance segmentationrobotic harvestingattachment inferenceocclusion rankingplant phenotypingsynthetic dataset
Authors
Rohit Menon, Niklas Mueller-Goldingen, Sicong Pan, Gokul Krishna Chenchani, Maren Bennewitz
Abstract
Robotic harvesting in dense crop canopies requires effective interventions that depend not only on geometry, but also on explicit, direction-conditioned relations identifying which organs obstruct a target fruit. We present SG-DOR (Scene Graphs with Direction-Conditioned Occlusion Reasoning), a relational framework that, given instance-segmented organ point clouds, infers a scene graph encoding physical attachments and direction-conditioned occlusion. We introduce an occlusion ranking task for retrieving and ranking candidate leaves for a target fruit and approach direction, and propose a direction-aware graph neural architecture with per-fruit leaf-set attention and union-level aggregation. Experiments on a multi-plant synthetic pepper dataset show improved occlusion prediction (F1=0.73, NDCG@3=0.85) and attachment inference (edge F1=0.83) over strong ablations, yielding a structured relational signal for downstream intervention planning.