Design, development, testing, and validation of robotic liquid application systems for precision agricultural applications

Abstract

The agricultural production system faces two major environmental issues: the overuse of pesticides and the accumulation of plastic waste. Traditional methods employ large self-propelled sprayers for applying herbicides, insecticides, and fungicides without adequate knowledge of the spatial variability of pests within the field, leading to overuse of pesticides. However, recent advancements in control strategies, including pulse width modulation (PWM) technology combined with artificial intelligence and robotics, enable precise spraying, thereby reducing pesticide use. Also, shifting from plastic mulch to sprayable biodegradable mulch aims to address plastic pollution in agricultural fields and promote sustainable farming practices. While traditional mulching involves laying sheets in the field, sprayable technology is gaining popularity due to its advantages. However, biodegradable mulch's properties and application volume provide a unique challenge, making the commercially available sprayer unsuitable for spraying. In the first study, a liquid application system was designed, developed, and integrated with an under-canopy rover for row crops. This system includes a tank, diaphragm pump, hollow cone nozzle, boom, and plumbing system. Two booms, each 55 inches in height, were mounted at the rear side of the rover, with each boom equipped with three nozzles and solenoid valves. The control system utilized an STM32 microcontroller, Electronic Control Unit (ECU), and drivers to implement PWM technology. The microcontroller was configured with the main computing platform using micro-ROS. The control system was validated by assessing duty cycle response and pressure variations with different nozzle configurations. The voltage remained in the excitation state for the specified duty cycles, and the system and nozzle pressure remained within ±5% of the target across different duty cycles and nozzle operations. Integration of the liquid application system with a pest detection system was achieved through a distance-based method using coordinate transformation. The 'tf2' library from ROS was used for this transformation, converting the detection frame from the camera to the world frame and then to the sprayer frame for precise spraying. The results demonstrated an overall effectiveness of 80%, with 100% accuracy in spraying plants infested with aphids. Additionally, 68.82% and 33.66% pesticide reductions were achieved, with pest severity levels of 11.11% and 22.22% compared to uniform whole-rate field spraying. Overall, the system's development and integration facilitated precise detection and spraying, significantly reducing pesticide usage. In the second study, a liquid application system was designed and developed for spraying biodegradable mulch. This system comprises a double-action positive displacement double diaphragm pump, a strainer, a pressure relief valve, flood jet nozzles, an agitation system, hoses, and fittings. A Sotera double diaphragm pump, capable of delivering a flow rate of 13 gallons per minute (GPM) at 13 PSI, was selected. Quick Turbo Flood Jet nozzles were chosen for their pre-orifice design, which minimizes drift and reduces clogging risk. The spraying system was mounted on the Amiga platform by FARM NG for its modularity and payload capacity. Experiments with various Quick Turbo Flood Jet (QCTF) nozzles identified the QCTF-60 nozzle as optimal for spraying the BioWRAP formulation, delivering 5.59 GPM at 10 PSI pressure and covering a 28-inch swath width. The rover's speed was set to 25 feet per minute to achieve the desired application volume of 4276 GPA. The sprayed BioWRAP formed a consistent layer across different soil types, performing best with higher moisture content and soil densities. The average thickness of the applied layer was 0.022 inches. This developed system will facilitate quick experimentation with the ongoing development of different biodegradable mulch formulations.