Design and Experiment of an Unoccupied Control System for a Tracked Grain Vehicle

Area of Science:

• Agricultural Engineering
• Robotics
• Computer Vision

Background:

• Enhancing crop harvesting efficiency is crucial for modern agriculture.
• Existing systems often lack full automation for grain transport and unloading.
• The need for autonomous agricultural machinery is increasing.

Purpose of the Study:

• To design and develop an automatic-driving tracked grain vehicle system.
• To integrate autonomous navigation and grain unloading functionalities.
• To evaluate the system's performance in field conditions.

Main Methods:

• Mechanical design of a 4.5 m³ tracked grain vehicle with a hydraulic unloading system.
• Electronic control system implementation using BODAS hydraulic controller for manual and automatic operation.
• Hybrid A* algorithm for path planning and a combined fuzzy-PID and pure pursuit algorithm for path tracking.
• Binocular vision and semantic segmentation for automatic grain unloading control (grain tank recognition and unloading regulation).

Main Results:

• Successful implementation of a hybrid manual/automatic electronic control system.
• Development of an automatic grain unloading control system utilizing computer vision.
• Field experiments demonstrated high precision in path-tracking control.
• Automatic unloading system met practical operational requirements.

Conclusions:

• The designed automatic-driving tracked grain vehicle system effectively enhances crop harvesting efficiency.
• The integrated navigation and unloading systems achieve the required precision for autonomous field operations.
• This technology offers a viable solution for practical unoccupied operations in agriculture.