Learning to Prune Branches in Modern Tree-Fruit Orchards

Jan 1, 2025·
Abhinav Jain
Abhinav Jain
,
Cindy Grimm
,
Stefan Lee
· 1 min read
Graphical Abstract: Our pipeline for pruning branch detection in modern orchards.
Abstract
Learning to Prune Branches in Modern Tree-Fruit Orchards
Type
Publication
2025 IEEE International Conference on Robotics and Automation (ICRA)
publications

Watch the Demo

See our robotic pruning system in action:

Abstract Details

Dormant tree pruning is labor intensive but essential to maintaining modern highly-productive fruit orchards. In this work we present a closed-loop visuomotor controller for robotic pruning. The controller guides the cutter through a cluttered tree environment to reach a specified cut point and ensures the cutters are perpendicular to the branch. We train the controller using a novel orchard simulation that captures the geometric distribution of branches in a target apple orchard configuration. Unlike traditional methods requiring full 3D reconstruction, our controller uses just optical flow images from a wrist-mounted camera. We deploy our learned policy in simulation and the real-world for an example V-Trellis envy tree with zero-shot transfer, achieving a ~30% success rate - approximately half the performance of an oracle planner.

Abhinav Jain
Authors
Abhinav Jain (he/him)
PhD Candidate in Robotics
Soon-to-be Ph.D. graduate in Robotics and AI researcher specializing in robotic mamipulation. Skilled in complex simulation design using Isaac Sim and Isaac Lab, and sim-to-real transfer for unstructured, real-world orchard environments. Expertise in creating novel RL algorithms, applying Deep Reinforcement Learning and Imitation Learning to deploy robust manipulation policies on physical robotic platforms. Recognized for research contributions in top-tier conferences (ICRA, CVPR, AAAI), with proven experience in synthetic data generation and training RL, generative AI, and Large Language Models (LLMs).