Vision-Based Leader Vehicle Trajectory Tracking for Multiple Agricultural Vehicles

Area of Science:

• Agricultural Engineering
• Robotics
• Computer Vision

Background:

• Efficient control of multiple vehicles in agriculture is challenging.
• Autonomous navigation systems can improve operational efficiency and reduce labor.
• Existing systems may lack accuracy or robustness in dynamic agricultural environments.

Purpose of the Study:

• To design and evaluate a leader-follower vehicle navigation system for agricultural applications.
• To enable an autonomous follower vehicle to accurately track a human-controlled leader vehicle.
• To enhance efficiency in agricultural operations through automated vehicle coordination.

Main Methods:

• Developed a monocular vision-based sensing system using a camera and rectangular markers.
• Implemented a least-squares method to reduce noise in data acquisition.
• Utilized a feedback control algorithm with a proportional-integral-derivative (PID) controller for trajectory tracking and distance maintenance.

Main Results:

• The leader-follower system demonstrated satisfactory tracking accuracy in field experiments.
• Root Mean Square (RMS) errors for linear trajectory tracking were as low as 6.5 cm.
• RMS errors for parallel trajectory tracking were as low as 7.1 cm.
• The autonomous follower vehicle successfully followed the leader vehicle at an average speed of 0.3 m/s.

Conclusions:

• The developed vision-based leader-follower system provides accurate and stable autonomous navigation.
• This technology is suitable for implementation in agricultural tasks such as grain harvesting.
• The system offers a practical solution for coordinating autonomous and human-operated vehicles in farming.