Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Abstract. Modern agriculture activities in the U.S. often require to be conducted over many fields with multiple vehicles cooperating together, even on a single farm. For example, during the harvesting season, combine harvesters, grain carts and trucks may need to work together to collect products over dozens of fields comprising a single farm. In this situation, it is difficult to capture and process what is exactly going on for all the vehicles and fields involved. However, to know what is happening is the first step for better management and performance improvement in these activities, which is especially important for the managers and the owners of the farm. In this paper, we propose a fully-automatic and easy-to-implement way to solve this problem. We have focused on the harvesting case and developed a rule-based algorithm to automatically recognize activities like harvesting and unloading, via GPS tracks. For evaluating the performance of the algorithm, we manually labeled GPS data collected over 10 fields during 2014 as the testing set. The overall accuracy is over 97% for detecting both harvesting and unloading activities. Hopefully, these key events identified by our algorithm can in addition make the product more traceable. We also developed some neural networks for solving the same problem, but with one simplified scenario. At the end of this paper, we will briefly discuss the challenges of utilizing neural networks, instead of a rule-based algorithm, for this problem.