Using the current prototype hardware, mount within a reasonable visual perspective of view of a sprayer nozzle and measure performance. Collect enough data to turn around a model in a short amount of time.
APPROACH
A current AI model for this did not exist. We mounted a camera to capture video and snapshots in increments of 5 seconds. This ran for several hours. Once finished, we compiled the data and sent it back to our data science team to begin model creation. Given the proprietary internal tooling we’ve built over the years, we had this model working by the next morning.
RESULTS
Customer advised that his father stated there were 80 degree tips on the sprayer. After Farmwave ran a spray pattern recognition, we found an average of 110 degrees per 10 frames. Customer, and father, went to check the sprayer and indeed Farmwave was correct; sprayer tips were at 110 degrees.
KEY LEARNINGS
In a short amount of time, additional data packages for the use of Farmwave can be created to perform additional tasks around the operation of equipment. We also recognize that current technology in sprayers does a good job of returning some results but we still saw an opportunity to improve. As an operator, it is tough to manage all aspects of the machinery as it moves through the field which is why we believe that displaying machinery performance incab based on specific thresholds set by the operator will vastly improve efficiencies and performance of the machinery. Below is a simple example how the Farmwave technology has been trained to measure nozzle performance. This technology will return results to the operator on the functionality of each of the nozzles on the boom ensuring proper flow and accurate coverage of the application.
