Implementation of a low-cost crop detection prototype for selective spraying in greenhouses

Abstract

During phytosanitary treatments aimed at minimizing product losses, selective spraying systems can be employed. These systems consist of a group of detection–action devices which manage the spraying. In this work, the technical feasibility of a low-cost ultrasound detection system prototype has been assessed for pesticide spray application on greenhouse crops. The prototype is based on a commercially-available car parking assistance system, which has been modified to amplify the signal and activate an electro-valve for spray control. This system was fitted into a self-propelled machine with two vertical spray booms. A laboratory test was carried out to evaluate the system limitations (detection range, response time, optimal sensor location); and once the feasibility of the system was known, a field test was conducted. Inside the greenhouse, the same parameters were determined for canopy presence. The system’s capacity to start and stop spraying at the beginnings and ends of the crop lines was also analysed. In addition, the minimum crop line surface with no plant mass that triggers system activation was determined. The results show that the detection range was 0–0.4 m with an average response time of 1.67 s. Based on these parameters, the optimal sensor location was determined for the different forward velocities. In conclusion, the results show that this system is suitable for plant detection at a forward speed of 0.9 m s−1, allowing growers to stop spraying automatically at the ends of the crop lines and where plant mass absence is greater than 1.0 lineal meter.