Development of an automated mechanical weed control system for intra-row weeds in vegetable row crops

Weed management is a tedious task especially in organic crop production since synthetic herbicides are prohibited, and manual weeding is relatively costly or unavailable. However, mechanical weed control is one of the options that can be implemented by growers to control weeds. Growers use simple implements to control weeds on their farms. Such tools are able to control weeds between crop rows, such as a basket weeder, and different type of sweeps. While there are numerous tools that offer excellent inter-row weed control, there are limited intra-row weed control methods.;The goal of this project was to develop and evaluate the performance of a prototype intra-row mechanical weed control system that is capable of controlling weeds in the intra-row area. The specific objectives were to design a hardware and software system to control the operation of the weeding mechanism in the intra-row area, and to determine the performance and accuracy of the crop detection system, and quantify physical crop damage.;Laboratory experiment to evaluate the accuracy of the crop detection system showed a linear relationship (P<.0001) between manually measured and sensor-estimated height. The model had an R2 of 0.93. The results showed that the ultrasonic sensor was able to detect the plant foliage (canopy) accurately with an estimated height of 30 cm compared with the manually measured height of 33 cm with standard deviation of 3 and 4.1, respectively.;The weeder ground speed and the plant intra-row distance were quantified in laboratory experiment. The speeds used in this test were 0.19 m/s (0.43 mph), and 0.58 m/s (1.30 mph). The crop plant intra-row distance was set based on the maximum distance between the two crop detection sensors.;The first field tests were to quantify the weed control efficacy for kale crop. The weeding efficacy was measured by counting the number of weeds in the centerline of a kale crop row before and after weeding. Results showed that the percent weed control was 41.7%.;The second field test was to quantify the potential weeded distance for two rows of sesame crop. The percentage of intra-row weeded distance for the first row was 34.4% while the percentage of the weeded distance for the second row was 49%.;A field test was conducted to determine the non-weeded distance for the intra-row weeder. The result was a graph between the weeding mechanism displacement (cm) and time (sec) for two plants. The non-weeded distances for the first plant were 20 cm before the weeding mechanism passes the plant and 29 cm after passing the plant. However, the non-weeded distances before and after passing the second plant were 18 cm and 18.5 cm, respectively.;A model was used to determine the non-weeded distance based on the plant height and the weeding mechanism distance and velocity. The model also gave the distance of crop detection system to the weeding mechanism.;The third field experiment was conducted to evaluate the weeded distance, and physical crop damage for bean and squash plants. The results showed that the average percent weeded distances for bean plants was 36.3%, while in squash it was 37.4%. There was no relationship between the distance between two plants and the percent weeded distance. The slope for the bean and squash crops were not significant at 5% level (P=0.36 and P=0.064) with slope standard error of 0.98 and 0.53, respectively.;The physical crop damage for bean and squash was quantified by counting the total number of leaves on each individual plant before and after weeding. The results showed that the average percent partial plant damage for beans was 20.3%, while for squash, the percent damage was 35%.;The fourth field experiment was conducted to evaluate the weed control efficacy. The results showed that there was no significant interaction among the main effects. However, there was a significant effect of weed species (P=0.03) on weed control efficacy. The Southern crabgrass with 33.9% was more effectively controlled than Florida pusley in which control was only 4.4%.