Real-time control of high-resolution micro-jet sprayer integrated with machine vision for precision weed control

摘要

The advent of automated technology in agriculture employing robots allows researchers and engineers to automate many of the tasks in a semi-structured, natural farming environment where these tasks need to be performed. Here we propose a fast-intelligent weed control system using a crop signalling concept with machine vision and a precision micro-jet sprayer to target in-row weeds for precision herbicide application. Crop signalling is a novel technology invented to read crop plants by machine to simplify the task of differentiating vegetable crops from weeds for selective weed control in real-time. In-row weed control in vegetable crops like lettuce requires a very precise herbicide spray resolution with a fast response time. A novel, accurate, high-speed, centimetre precision spray targeting actuator system was designed and experimentally validated in synchronization with a machine vision system to spray detected weeds located between lettuce plants. The system processed an image, representing a 120 mm × 180 mm region of row-crop in 80 ms, which allowed the micro-jet sprayer to successfully function at a travel speed of 3.2 km h−1 and selectively deliver herbicide to the weed targets. The analysis of the overall performance of the system to kill weeds in indoor experimental trials is discussed and presented. Findings indicate that 98% weeds were correctly sprayed which indicates the efficacy and robustness of the proposed systems.