Robotic tomato field mapping
2026
← All Research
AR Lab / Research
AI-enabled Plant Phenotyping
We develop robotic sensing systems and computer vision methods to measure plant traits, map crop production, and characterize plants in indoor and field environments. Our projects span robotic imaging, high-throughput phenotyping, yield mapping, and 3D reconstruction.
Project Videos
High-throughput Robotic Phenotyping
2026
Yield Mapping with Severe Occlusions in the Field
2026
BenchBot
2024
Amiga Robot
2023
Robotic Leaf Angle Characterization
2019 - 2022
AngleNet is an automated image processing pipeline developed for measuring leaf angles of maize plants in the field. This pipeline detects each leaf angle as a triplet of keypoints in two-dimensional images and extracts quantitative data from reconstructed 3D models.
3D Model of Sorghum Stems
2020
Our customized PhenoStereo cameras are capable of capturing high-quality and high sharpness stereoscopic images under field conditions. The images are used to reconstruct high-fidelity 3D models by stereo matching algorithms.
3D Model of Maize Plants
2021
The multiview 3D point clouds were cross-registered to reconstruct the entire canopy of maize plants in the field.
Related Publications
- Robotic leaf angle characterization / AngleNet
- PhenoStereo / 3D reconstruction of sorghum stems
- PD-CLIP: A contrastive language-image pre-training framework for zero-shot fine-grained plant disease diagnosis
- Swin-Roleaf: A new method for characterizing leaf azimuth angle in large-scale maize plants
- High‐Throughput Robotic Phenotyping for QuantifyingTomato Disease Severity Enabled by Synthetic Data andDomain‐Adaptive Semantic Segmentation