Autonomous Robotic System to Prune Sweet Pepper Leaves Using Semantic Segmentation with Deep Learning and Articulated...

Area of Science:

• Agricultural Robotics
• Computer Vision
• Deep Learning

Background:

• Automated agricultural tasks are crucial for efficiency and sustainability.
• Precision agriculture requires sophisticated robotic systems for plant management.
• Sweet pepper cultivation can benefit from automated pruning for yield optimization.

Purpose of the Study:

• To develop an autonomous robotic system for pruning sweet pepper leaves.
• To integrate semantic segmentation, 3D point cloud analysis, and manipulator control for pruning.
• To demonstrate the feasibility of robotic pruning in a controlled agricultural environment.

Main Methods:

• Utilized deep learning-based semantic segmentation to identify sweet pepper plant parts.
• Generated 3D point clouds from depth camera data for precise pruning position detection.
• Implemented an articulated manipulator robot controlled via the Robot Operating System (ROS).

Main Results:

• Successfully recognized different parts of the sweet pepper plant using semantic segmentation.
• Accurately detected pruning positions and determined manipulator poses from 3D point clouds.
• Demonstrated robotic pruning of sweet pepper leaves within a specified height range.

Conclusions:

• The proposed autonomous system effectively integrates perception and control for robotic pruning.
• Semantic segmentation and 3D point cloud analysis are key components for precise agricultural robotics.
• This system offers a foundation for advanced automation in horticultural practices.