Development of Machine Vision Node on Standardized Controller Area Network of a Self-Propelled Boom Sprayer to Perform Site-Specific Pesticide Application

Abstract

Pesticides are applied by commercial sprayers to control weeds and insects, which commonly develop in patches. Several boom sprayers are equipped with ISOBUS communication links which is a common tendency to integrate different devices and improve the robustness and efficiency of agriculture machineries. The ISOBUS uses CAN-bus as a data link protocol to perform the communication. Machine vision systems (MVS) and real-time response spraying have the potential to limit pesticide use by only spraying on infested areas. However, difficulties arise with integrating MVS with sprayers. The goal of this study is to develop a machine vision node (MVN) for MV-based site-specific spraying on ISOBUS-compatible sprayers. The MVN acts as an ECU which reads and sends the detected information from the MVS to control nozzles. The sprayer apparatus consists of a virtual terminal, a hydraulic unit, a nozzle control unit, a nozzle electronic switch, and six nozzles. A communication algorithm was developed on MVN to read the detected results from the MVS and send respective communication messages via serial communication to the sprayer. The developed algorithm was tested and evaluated with static (small apparatus including six nozzles) and commercial ISOBUS-compatible sprayers to control each nozzle. The results show that correct spraying commands were sent to nozzles (64 combinations) without data loss between the MVS and sprayer. Overall, the developed system resulted in a timely decision (10 ms) for site-specific spraying which may enable real-time spot spraying using machine vision.