VineSLAM

A multi-layer Localization and Mapping procedure for agricultural places.

Supported by ROSIN - ROS-Industrial Quality-Assured Robot Software Components.
More information: rosin-project.eu

This project has received funding from the European Union’s Horizon 2020
research and innovation programme under grant agreement no. 732287.

Table of Contents

1. System architecture
2. System components
3. Installation
4. ROS structure
5. How to run

System architecture

VineSLAM relies on three main software blocks:

• Feature extractor: computes high-level semantic features, geometric features, and image features.
• Multi-layer map: considers all the extracted features to build individual maps that together constitute a novel 3D multi-layer map.
• Localization: Particle filter that uses the feature extractor and the multi-layer map to localize the robot.

System components

---

Semantic Map

Uses a deep Learning-based object detector to capture semantic features from the agricultural environment.

• Hardware requirements: Google's Coral USB Accelerator

Example of vine trunk detection

Figueiredo Vineyard	Aveleda Vineyard	Aveleda Vineyard Thermal

Geometric maps

A 3D map with features - corners, planars and semi-planes directly extracted from a 3D point cloud.

Watch the following videos to see the multi-layer map creation, regarding geometric maps.

• Corner features maps

• Planar features maps

• Semi-plane features maps

Elevation Map

Elevation map extracted from the ground plane. It considers a well-defined grid map structure with altitude information for each cell.

Visual maps

A 3D map built with SURF features detected on the image.

Installation

sudo apt-get install ros-foxy-vision-msgs ros-foxy-shape-msgs

cd <path_to_ros2_ws>/src
git clone git@gitlab.inesctec.pt:agrob/vineslam.git -b master
cd <path_to_ros2_ws>
colcon build --cmake-args -DCMAKE_BUILD_TYPE=Release

ROS structure

VineSLAM requires many input topics from sensor data, and outputs many topics (robot pose and maps). Also, some parameters can be tuned to improve its performance. All this information is summarized in the configuration file description.

How to run

• You can use the already implemented launch files to each node.
• If you want to use your own topics, just edit them and remap the topics.
• SLAM node:

roslaunch vineslam_ros test_slam.launch

• Localization node:

roslaunch vineslam_ros test_localization.launch