VinePT-Map: Pole-Trunk Semantic Mapping for Resilient Autonomous Robotics in Vineyards

Giorgio Audrito; Mauro Martini; Alessandro Navone; Giorgia Galluzzo; Marcello Chiaberge

cs.ROMar 5, 2026

VinePT-Map: Pole-Trunk Semantic Mapping for Resilient Autonomous Robotics in Vineyards

Authors: Giorgio Audrito, Mauro Martini, Alessandro Navone, Giorgia Galluzzo, Marcello Chiaberge

Organizations: Department of Electronics and Telecommunications, Politecnico di Torino, 10129, Torino, Italy.

Abstract

Reliable long-term deployment of autonomous robots in agricultural environments remains challenging due to perceptual aliasing, seasonal variability, and the dynamic nature of crop canopies. Vineyards, characterized by repetitive row structures and significant visual changes across phenological stages, represent a pivotal field challenge, limiting the robustness of conventional feature-based localization and mapping approaches. This paper introduces VinePT-Map, a semantic mapping framework that leverages vine trunks and support poles as persistent structural landmarks to enable season-agnostic and resilient robot localization. The proposed method formulates the mapping problem as a factor graph, integrating GPS, IMU, and RGB-D observations through robust geometrical constraints that exploit vineyard structure. An efficient perception pipeline based on instance segmentation and tracking, combined with a clustering filter for outlier rejection and pose refinement, enables accurate landmark detection using low-cost sensors and onboard computation. To validate the pipeline, we present a multi-season dataset for trunk and pole segmentation and tracking. Extensive field experiments conducted across diverse seasons demonstrate the robustness and accuracy of the proposed approach, highlighting its suitability for long-term autonomous operation in agricultural environments.

Figures & tables

Fig. 1: VinePT-Map enables robotic platforms to build permanent landmarks map in vineyards for robust autonomous operation in all the seasons.

Fig. 2: VinePT-Map: a schematic pipeline of the semantic mapping framework for persistent landmarks in vineyards.

Fig. 3: VinePT-Map Factor Graph for poles and trunks mapping.

Campaign	Weather	Plant Growth	Grass Height
12/02	Cloudy	Stem, no leaves	Low ( $\sim 5$ cm)
20/03	Sunny	Stem, no leaves	Medium ( $\sim 20$ cm)
14/08	Sunny	Branches, leaves, fruit	Low ( $\sim 5$ cm)
11/09	Cloudy	Branches, leaves, fruit	Medium ( $\sim 20$ cm)

TABLE I: Overview of field campaigns in the vineyards.

Fig. 4: Aerial view of the vineyard with the trajectory of the robot over three rows (starting from dark line, ending in yellow line), used to compute the incremental result of Table III .

Class	Date	Box mAP50	Mask mAP50	HOTA	AssA
	02-12	0.934	0.928	89.1	90.3
	03-20	0.961	0.946	85.4	89.8
Trunks	08-14	0.91	0.916	74.7	87.1
	09-11	0.863	0.859	81.3	84.8
	Overall	0.914	0.908	82.6	88.0
	02-12	0.878	0.833	72.4	74.7

TABLE II: Results of landmarks instance segmentation and tracking. The best seasonal results are highlighted in bold.

N. Rows	1		2		3
Date	MAE [m] $\downarrow$	TP $(\%)\uparrow$	MAE [m] $\downarrow$	TP $(\%)\uparrow$	MAE [m] $\downarrow$	TP $(\%)\uparrow$
02-11	0.20	100.0	0.18	97.3	0.18	97.1
03-20	0.24	96.2	0.22	96.3	0.27	91.1
08-14	0.25	98.1	0.24	97.2	0.24	93.1
09-11	0.31	100	0.29	96.3	0.26	97.2

TABLE III: Quantitative evaluation of poles landmarks mapping using georeferenced positions as ground truth. The metrics are reported incrementally over three consecutive rows.

Fig. 5: Box plots of the ablation study results on poles mapping error. The relevance of the landmark reference point computation method (rd) and the data association method (rc) is highlighted by excluding them from the pipeline.

Panel of Figure 9 — Fig. 6: Map of vineyard poles and trunks over three consecutive rows in February, March, August, and September, with a zoomed-in view of the poles and trunks.

Explore similar work

An Annotation-to-Detection Framework for Autonomous and Robust Vine Trunk Localization in the Field by Mobile Agricultural Robots
Mar 20, 2026Dimitrios Chatziparaschis, Elia Scudiero, Brent Sams +1Object Detection Multimodal Sensor Fusion
Dept. of Electrical and Computer Engineering, Univ. of California, Riverside, 900 University Avenue, Riverside, CA 92521, USA · Dept. of Environmental Sciences, Univ. of California, Riverside, 900 University Avenue, Riverside, CA 92521, USA · Gallo, 600 Yosemite Blvd, Modesto, CA 95354, USA
Semantic Landmark Particle Filter for Robot Localisation in Vineyards
Mar 11, 2026Rajitha de Silva, Jonathan Cox, James R. Heselden +3Robot Localization LiDAR Localization
Lincoln Centre for Autonomous Systems (L-CAS), University of Lincoln, UK. · MAVLab, TU Delft, Netherlands. · Robotics Systems Lab, ETH Zurich, Switzerland.
Semantic SLAM in Precision Agriculture using Bayesian Inference
Sep 17, 2026Ruben Beumer, Sander Doodeman, René van de Molengraft +1Simultaneous Localization and Mapping Semantic Mapping
Department of Mechanical Engineering, Eindhoven University of Technology P.O. Box 513, 5600 MB Eindhoven, The Netherlands

Mar 20, 2026J/K move · Enter open · S save

An Annotation-to-Detection Framework for Autonomous and Robust Vine Trunk Localization in the Field by Mobile Agricultural Robots

Dimitrios Chatziparaschis, Elia Scudiero, Brent Sams, Konstantinos Karydis

Dept. of Electrical and Computer Engineering, Univ. of California, Riverside, 900 University Avenue, Riverside, CA 92521, USA · Dept. of Environmental Sciences, Univ. of California, Riverside, 900 University Avenue, Riverside, CA 92521, USA · Gallo, 600 Yosemite Blvd, Modesto, CA 95354, USA

Object Detection Multimodal Sensor Fusion LiDAR Localization

Open paper AskPDF arXiv

The dynamic and heterogeneous nature of agricultural fields presents significant challenges for object detection and localization, particularly for autonomous mobile robots that are tasked with surveying previously unseen unstructured environments. Concurrently, there is a growing need for real-time detection systems that do not depend on large-scale manually labeled real-world datasets. In this work, we introduce a comprehensive annotation-to-detection framework designed to train a robust multi-modal detector using limited and partially labeled training data. The proposed methodology incorporates cross-modal annotation transfer and an early-stage sensor fusion pipeline, which, in conjunction with a multi-stage detection architecture, effectively trains and enhances the system's multi-modal detection capabilities. The effectiveness of the framework was demonstrated through vine trunk detection in novel vineyard settings that featured diverse lighting conditions and varying crop densities to validate performance. When integrated with a customized multi-modal LiDAR and Odometry Mapping (LOAM) algorithm and a tree association module, the system demonstrated high-performance trunk localization, successfully identifying over 70% of trees in a single traversal with a mean distance error of less than 0.37m. The results reveal that by leveraging multi-modal, incremental-stage annotation and training, the proposed framework achieves robust detection performance regardless of limited starting annotations, showcasing its potential for real-world and near-ground agricultural applications.

Abstract

Explore similar work

An Annotation-to-Detection Framework for Autonomous and Robust Vine Trunk Localization in the Field by Mobile Agricultural Robots

Semantic Landmark Particle Filter for Robot Localisation in Vineyards

Semantic SLAM in Precision Agriculture using Bayesian Inference