An Automated System for Crop Signaling and Robotic Weed Control in Processing Tomatoes

Abstract. Weed control for vegetable crops in the United States is predominantly done with either herbicides or manual labor. Due to a combination of weeds developing resistance to some herbicides and consumers demanding vegetables not treated with conventional chemicals, herbicides are becoming less appealing as a weed control tool. Manual labor for hoeing weeds by hand continues to be widely used, however research has shown that hand-hoeing crews mistake weeds for crop plants or miss weeds and eliminate only about 65 to 85% of the weeds on average. Hand weeding is also plagued by a lack of available labor and high cost. For these reasons, farmers are interested in the development of an automated weed control system that can effectively remove weeds growing among crop plants. In a recent review of robotic weed control technologies our group has identified reliable crop vs. weed discrimination as the principal technological challenge to the commercialization of robotic weeding machines for high weed density fields.

Our group at UC Davis has developed a novel solution to the weed discrimination problem using a technique called Crop Signaling. In Crop Signaling, the crop plants are treated with a unique machine-readable signaling compound at planting. The crop signal is then used during the remainder of the season to allow smart machines to automatically care for crop plants including weed control tasks. This paper reports on the design and evaluation of an automatic system for application of the Crop Signaling compound to vegetable plants during transplanting.