STEMbot: A Compliant Robot for Under-Canopy Plant Navigation

Jul 16, 2026

Organic agriculture scalability is limited by labor-intensive pest monitoring: many pests are detectable only on leaf undersides or stems after significant damage. STEMbot presents a miniature climbing robot system for autonomous navigation beneath plant canopies. It combines a geometric PIN-SLAM pipeline with a semantic OcTree for robust localization and mapping during climbing. A manifold-constrained A* planner and ray-tracing goal specification enable branch-aware traversal and occluded target inspection, validated by hardware experiments on diverse specimens.

LanguageEnglish

CategoryTechnology

FormatPDF

Pages8

Reading Time20 min

Introduction

Related Work

Climbing Robots and Tree Traversal

Contributions

STEMbot: A Compliant Robot for Under-Canopy Plant Navigation Zachary Charlick1 , Nilay Roy Choudhury 1 , Haoyu Ma 1 , Xiaonan Huang 1 , and Dmitry Berenson 1 arXiv :2607 .07873v 1 [ cs .RO] 8 Jul 2026 Abstract—The scalability of organic agriculture is partially limited by the labor costs associated with monitoring for pests. While drones and rovers are well-suited for agricultural monitoring from above or next to plants, many pests live on the underside of leaves or on plant stems, making them detectable only after they have caused significant damage. To enable early pest detection we present STEMbot, a miniature climbing robot system designed for autonomous navigation under plant canopies. Unlike existing climbing platforms that lack on-board perception or are restricted to unbranched vertical trunks, STEMbot integrates a fully geometric PIN-SLAM pipeline with a semantic OcTree to achieve robust localization and mapping while climbing the plant. To plan STEMbot’s motion we propose a manifold-constrained A* planner along with ray-tracing goal specification to enable branch-aware traversal and the inspection of occluded targets. We validate our system through hardware experiments, demonstrating reliable traversal of stems ranging from 7–33 mm and autonomous navigation across four distinct plant specimens. Quantitative evaluations show that our system achieves high-fidelity geometric reconstructions with an average Chamfer distance of less than 1 cm relative to an offline photogrammetry baseline, confirming that STEMbot maintains the globally consistent odometry needed for autonomous navigation. I. INTRODUCTION Existing methods for insect pest detection in organic vegetable farming—e.g. Integrated Pest Management (IPM) [1]—involve frequent manual inspections and pest-specific trapping at key times during a pest’s lifecycle. These methods require significant labor and expertise, and the associated expense limits the scalability of organic farming. While dronesand rovers are able to monitor plants from a distance, many pests live on the underside of leaves or on plant stems, making them detectable from a distance only after they have caused significant damage. In this systems paper, we propose STEMbot, a novel climbing robot design and software framework for autonomous navigation under plant canopies. By navigating on stems and branches, this robot is able to work around occlusions so that it can inspect specific areas of the plant where a target pest is likely to reside* . Automating plant navigation requires robust perception, that is able to handle unstructured geometry and extreme occlusion. Furthermore, the repetitive textures and largely monochromatic appearance of plant foliage introduce sub- 1Robotics Department, University of Michigan, Ann Arbor, MI, USA. 2This work was supported in part by the Office of Naval Research Grant N00014-24-1-2036, NSF grants IIS-2113401 and IIS-2220876, and the National Robotics Initiative, project award no. 2026-67021-46039, from the U.S. Department of Agriculture’s National Institute of Food and Agriculture. * Specifically, this robot is intended for high-value crops with climbable stems such as tomatoes, peppers, and cucurbits to detect pests such as tomato hornworms, cucumber beetles, and aphids. Fig. 1. STEMbot navigating under a plant canopy. The robot integrates perception, motion planning, and specialized hardware to maintain stable contact with the stem manifold, enabling autonomous localization and semantic mapping. stantial perceptual aliasing, which degrades the performance of many perception methods. While many climbing robots have been developed, most are designed for industrial inspection tasks, such as robots that climb and inspect pipes or infrastructure surfaces [2], [3],[4], [5] . Tree-climbing robots have also been explored, but many of these systems are designed to fully-enclose a single trunk, such as a coconut tree, where the absence of branch

FAQs

What problem does STEMbot target in organic agriculture pest monitoring?

STEMbot targets the high labor cost of monitoring pests, since many pests live on leaf undersides or plant stems and are only detectable after damage. It enables earlier detection through autonomous inspection under canopies.

How does STEMbot achieve robust localization and mapping while climbing?

STEMbot integrates a geometric PIN-SLAM pipeline with a semantic OcTree. This combination supports robust localization and mapping despite occlusion during plant traversal.

What motion planning approach does STEMbot use for branch-aware navigation?

STEMbot proposes a manifold-constrained A* planner with ray-tracing goal specification. This supports navigation that accounts for plant branches and occluded targets.

Jul 16, 2026

Upload by Clementine