TOWARDS DEVELOPING STANDARDIZED PRECISION AGRICULTURE BOOM SPRAYER VIA HYBRID COMMUNICATION NETWORK FOR REAL-TIME SPOT APPLICATION

Abstract

The transition of boom spraying towards spot application under precision agriculture schemes faces challenges due to the large volume of data generated by a large number of sensing and actuation devices. This research focuses on developing a universal communication network for real-time spot application, using Controller Area Network (CAN) at its core, offering the advantages of potentially error-free communication and seamless integration of machine vision systems into different boom sprayers. To handle the narrow bandwidth characteristic of CAN, a novel electronic control unit (ECU) was developed to encapsulate pest detection results into CAN data frames based on detected pest locations in images received from one machine vision system consisting of multiple cameras. The machine vision data were transmitted through UART to identify the number of nozzles to be actuated via CAN. The ECU was designed to accommodate different machine vision systems with varying camera counts and image resolutions. For real-time control, the ECU extracted data every 40 ms and constructed CAN frames in two separate threads simultaneously. Field tests demonstrated that the ECU managed nozzle actuation for targets distributed across diverse scenarios, including spatial and temporal successions. Since the conditions on wide boom sprayers require multiple machine vision systems to actuate dozens of nozzles, an upgraded communication protocol was built at the interface of the machine vision with the ECU based on Ethernet. An application layer based on ISO 11783 was added to the CAN interface, widely used in agricultural machinery including sprayers. These upgrades allowed handling nozzle actuation at variable sprayer speeds up to 9.66 kph with a minimum spray length of 345 mm per detection, processing over 30 data frames every 40 ms. Finally, a new ISO 11783-compliant CAN bus with 60 nozzles was installed on a 36 m boom sprayer, used as a case study. This new bus featured two additional ECUs: one to communicate with other buses in the sprayer to import data like speed, and another to store pest detection and nozzle actuation data for further analysis. The case study demonstrated that a complete real-time spot application mechanism, including 30 cameras, would require an additional 4034 W for full functionality.