The development and assessment of the accuracy of an autonomous GPS-based system for intra-row mechanical weed control in row crops

Highly selective mechanical weed control within crop rows as an alternative to herbicide treatment requires the accurate guidance of tools around individual crop plants. The objectives of the research presented in this paper were to develop and optimise a novel self-propelled and unmanned hoeing system for intra-row weed control based on real-time kinematic global positioning system (RTKGPS) navigation, and to evaluate its performance under field conditions. The system comprised an autonomous tractor and a side-shifting frame, both equipped with RTK-GPS, and an attached cycloid hoe, which is a tine-rotor with eight sigmoid-shaped, vertically directed tines that can be released to allow individual rotation in order to avoid collision with geo-referenced crop plants. The system navigates with reference to predefined waypoints for hoeing parallel to crop rows and around individual crop plants. Field experiments using plastic sticks as artificial crop plants, placed with a mean within-row spacing of 0.2 m and variation typical under field conditions, confirmed that the system reliably performed hoeing within crop rows without colliding with the plastic sticks at forward velocities up to 0.52 m s−1. The side-shift enabled control of the transverse position of the cycloid hoe and was able to follow the row line with an accuracy of ±16 mm (P = 0.95) at 0.31 m s−1 and ±22 mm (P = 0.95) at 0.52 m s−1 forward velocity. Tines intruded into the required uncultivated zone, which had a radius of 10 mm from the centre of individual plants, with a maximum distance of 9 mm for 18 out of 1224 observations. Sufficient accuracy of the real-time positioning and control system was provided by using a Kalman filter, signal processing of the tilt data, and control algorithms developed for the side-shift and cycloid hoe control systems.