VISION-BASED AGRICULTURAL IMPLEMENT AWARENESS SYSTEM

Abstract

In current agricultural technology, most tractors come furnished with automation capabilities. The implements that a tractor can pull, however (grain cart, baler, planter, etc..), are not automated, and cannot be simply hitched to the tractor without the system being able to keep track of the machine that it is pulling. Our system aims to make that functionality a possibility, while centralizing the automation technology on just the tractor itself, by developing a visual sensor system that can track important information about the implement while the tractor is operating. We implemented this system as a proof-of-design using a standard pick-up truck and U-Haul Trailers. Some of the most important requirements of our system is to recognize the type of trailer the truck is pulling with 80% accuracy, determine the angle at which the trailer is currently hitched to the truck within a 10⁰ margin, and calculate the position of the trailer relative to the truck within 15cm, all while running off 12V DC. Our design was able to meet all of these key system requirements, as verified through testing. This system was developed as a UArizona Engineering Senior Capstone project, our team number was 23046, our sponsor was AGCO, and our mentor was Jim Sweetman.