RICE: Reactive Interaction Controller for Cluttered Canopy Environment
Nidhi Parayil, Thierry Peynot, Christopher Lehnert
AI summary
Problem
Robotic manipulation in dense agricultural canopies is hindered by physical and visual occlusion, causing traditional vision-based or model-dependent controllers to fail when navigating around or through delicate, deformable obstacles without causing damage.
Approach
The authors propose RICE, a model-free hierarchical controller that continuously balances goal-directed motion with tactile force feedback to adaptively choose between pushing through or maneuvering around deformable obstacles in real time.
Key results
- Successfully reached occluded targets in all 35+ trials across three plant setups without breaking branches
- Outperformed established hybrid and position-based control strategies in reliability and adaptability
- Introduced a quantitative evaluation framework using motion capture to measure branch deformation during interaction
- Demonstrated robust performance in both structured mock plants and unstructured artificial canopies
Why it matters
It enables reliable, damage-free robotic manipulation in unstructured agricultural environments, paving the way for automated harvesting and pruning tasks.
Abstract
Robotic motion in dense, cluttered environments such as agricultural canopies presents significant challenges due to physical and visual occlusion caused by leaves and branches. Traditional vision-based or model-dependent ap- proaches often fail in these settings, where physical interaction without damaging foliage and branches is necessary to reach a target. We present a novel reactive controller that enables safe navigation for a robotic arm in a contact-rich, cluttered, deformable environment using end-effector position and real- time tactile feedback. Our proposed frameworkâs interaction strategy is based on a trade-off between minimizing disturbance by maneuvering around obstacles and pushing through them to move towards the target. We show that over 35 trials in 3 experimental plant setups with an occluded target, the proposed controller successfully reached the target in all trials without breaking any branches and outperformed the established con- trol strategy for dense foliage in reliability and adaptability. This work lays the foundation for safe, adaptive interaction in cluttered, contact-rich deformable environments, enabling future agricultural tasks such as pruning and harvesting in plant canopies.